I, the undersigned, of c/o SAKAI International Patent Office, 8-1, Kasumigaseki 3-chome, Chiyoda-ku, Tokyo 100-0013 Japan, hereby declare that I am a translator of the document attached, and attached document is a true and correct translation made by me to the best of my knowledge and belief. PCT Application No.PCT/JP2015/073055, filed on Aug. 17, 2015 Signature of Translator: jVt^ / ' ^ ' ^twu Kazuyuki KAWAHARA Date: Mar. 1, 2017 Docket No.: PMAA-15086-PCT 1 DESCRIPTION CONTROL APPARATUS AND CONTROL METHOD 5 Field [0001] The present invention relates to a control apparatus and a control method for performing control of a target apparatus according to feedback control. 10 Background [0002] In an industrial apparatus, feedback control is used to realize necessary operation. In the feedback control, adjustment of a control gain used to calculate an operation amount from a detected control amount is 15 necessary. [0003] As one of methods for appropriately adjusting the control gain, there is a limit cycle method. The limit cycle method is a method of, by performing control for selecting one from two operation amounts and outputting the 20 operation amount called binary control, vibrating a control amount at a constant cycle or a cycle regarded as constant, identifying dynamic characteristics of a control target on the basis of a vibration waveform, and calculating a control gain (Non Patent Literature 1). The vibration 25 waveform is called limit cycle waveform. [0004] In the limit cycle method, to identify the dynamic characteristics of the control target, it is necessary to select one of the operation amounts of the binary control on the basis of a plus or minus sign of a 30 control deviation, vibrate the control amount at the constant cycle or the cycle regarded as constant, and generate the limit cycle waveform. However, in the limit cycle method, when noise is included in the detected Docket No.: PMAA-15086-PCT 2 control amount and the plus or minus sign of the control deviation is reversed because of the influence of the noise, on/off control operates irrespective of the dynamic characteristics of the control target. In some case, 5 chattering occurs, the limit cycle waveform of the constant cycle or the cycle regarded as constant cannot be obtained, and the dynamic characteristics of the control target cannot be identified. [0005] Patent Literature 1 mentions that, even when 10 noise is included in the detected control amount, the limit cycle waveform of the constant cycle or the cycle regarded as constant is generated. Patent Literature 1 mentions that the chattering of the binary control due to the influence of the noise included in the detected control 15 amount is prevented by providing a hysteresis characteristic in the determination of the plus or minus sign of the control deviation. Patent Literature 1 mentions that the dynamic characteristics of the control target is identified and the control gain is calculated 20 even when noise is included in the detected control amount. Citation List Patent Literature [0006] Patent Literature 1: Japanese Patent Application 25 Laid-Open No. S61-279901 Non Patent Literature [0007] Non Patent Literature 1: Nobuhide Suda, "PID control", System Control Information Library, Asakura Publishing Co., Ltd., pp 162 to 167 30 Summary Technical Problem [0008] However, in the technology disclosed in Patent Docket No.: PMAA-15086-PCT 3 Literature 1, unless the magnitude of hysteresis of the hysteresis characteristic provided in the determination of the plus or minus sign of the control deviation is appropriately set, an accurate limit cycle waveform cannot 5 be obtained. It is difficult to accurately perform the calculation of the control gain. [0009] When the magnitude of set hysteresis is small and inappropriate, the influence of the noise included in the detected control amount cannot be eliminated. The 10 chattering occurs because the binary control operates irrespective of the dynamic characteristics of the control target. The limit cycle waveform of the constant cycle or the cycle regarded as constant cannot be obtained. The calculation of the control gain cannot be accurately 15 performed. [0010] When the magnitude of the set hysteresis is too large and inappropriate, even if the influence of the noise can be eliminated, it is likely that the vibration amplitude of the control amount increase, the control 20 amount is saturated and the limit cycle waveform changes to a shape unrelated to the dynamic characteristics of the control target, and the control target is damaged. The calculation of the control gain cannot be accurately performed. 25 [0011] Therefore, when the limit cycle method is carried out on the basis of Patent Literature 1, an operator needs to observe beforehand the magnitude of the noise included in the detected control amount and set the magnitude of the hysteresis to a degree capable of eliminating the influence 30 of the noise and not excessively increasing the vibration amplitude of the control amount. Therefore, Patent Literature 1 has a problem in that considerable labor and time are required to set appropriate magnitude of the Docket No.: PMAA-15086-PCT 4 hysteresis. [0012] In the case of a configuration in which the user sets the magnitude of the hysteresis, it is necessary to provide, in a control apparatus, a function for displaying 5 a control amount such that the operator can observe the control amount in detail from the outside and a function for setting the magnitude of the hysteresis from the outside. Therefore, there is also a problem in that Patent Literature 1 increases production cost of the control 10 apparatus. [0013] Further, in a specific operation environment, even if the magnitude of the hysteresis is appropriately set by the operator, the magnitude of the noise included in the control amount increases according to a change in the 15 magnitude of the control amount, and the magnitude of the noise included in the control amount increases according to an operation state of an electric machine or an electronic apparatus set around the control apparatus and the control target. In some case, the influence of the noise cannot be 20 eliminated and the calculation of the control gain cannot be accurately performed. In such a case, the operator needs to change the magnitude of the hysteresis on the basis of the magnitude of the control amount or on the basis of the operation state of the electric machine or the 25 electronic apparatus set around the control apparatus and the control target. The operator also needs to check the magnitude of the noise included in the control amount. Therefore, Patent Literature 1 also has a problem in that the operator needs to continue to change the setting of the 30 magnitude of the hysteresis and considerable time and labor are required. [0014] The present invention has been devised in view of the above and an object of the present invention is to Docket No.: PMAA-15086-PCT 5 provide a control apparatus and a control method that can eliminate the influence of noise and accurately calculate a control gain by appropriately setting the magnitude of hysteresis. 5 Solution to Problem [0015] In order to solve the aforementioned problem and achieve the object, the present invention provides a control apparatus including: a subtracter that calculates a 10 control deviation by subtracting a command value input from an outside and a control amount input from a control target apparatus; a control computing unit that generates an operation amount on the basis of the control deviation and a control gain and outputs the operation amount; an 15 adjustment-execution-command generating unit that generates an adjustment-execution command value indicating ON or OFF and outputs the adjustment-execution command value; a binary output unit that generates, in a period in which the adjustment-execution command value output from the 20 adjustment-execution-command generating unit is ON, an adjustment-time addition value on the basis of the control deviation and a hysteresis-width setting value and outputs the adjustment-time addition value; a standard-deviation estimating unit that calculates, in a period in which the 25 adjustment-execution command value output from the adjustment-execution-command generating unit is OFF, a low-frequency-component removed signal obtained by removing low-frequency components of the control amount or the control deviation and calculates a standard-deviation 30 estimated value, which is an estimated value of a standard deviation of the low-frequency component removed signal; and a hysteresis-width computing unit that calculates a hysteresis-width computed value on the basis of the Docket No.: PMAA-15086-PCT 6 standard-deviation estimated value and changes the hysteresis-width setting value of the binary output unit to the hysteresis-width computed value. 5 Advantageous Effects of Invention [0016] According to the present invention, there is an effect that the influence of noise is eliminated and a control gain is accurately calculated by appropriately setting the magnitude of hysteresis. 10 Brief Description of Drawings [0017] FIG. 1 is a configuration diagram of a control apparatus and a control target apparatus according to a first embodiment. 15 FIG. 2 is a diagram schematically showing the operation of a binary output unit according to the first embodiment. FIG. 3 is a configuration diagram of a standard-deviation estimating unit according to the first embodiment 20 FIG. 4 is a configuration diagram of a control apparatus and a control target apparatus according to a second embodiment. FIG. 5 is a configuration diagram of a standard-deviation estimating unit according to the second 2 5 embodiment. FIG. 6 is a configuration diagram of a control apparatus and a control target apparatus according to a third embodiment. FIG. 7 is a configuration diagram of a standard-30 deviation estimating unit according to the third embodiment FIG. 8 is a configuration diagram of a control apparatus and a control target apparatus according to a fourth embodiment. Docket No.: PMAA-15086-PCT 7 FIG. 9 is a configuration diagram of a standard-deviation estimating unit according to the fourth embodiment. FIG. 10 is a configuration diagram of a control 5 apparatus and a control target apparatus according to a fifth embodiment. FIG. 11 is a configuration diagram of a standard-deviation estimating unit according to the fifth embodiment FIG. 12 is a configuration diagram of a control 10 apparatus and a control target apparatus according to a sixth embodiment. Description of Embodiments [0018] Control apparatuses according to embodiments of 15 the present invention are explained in detail below with reference to the drawings. Note that the present invention is not limited by the embodiments. [0019] First Embodiment. FIG. 1 is a configuration diagram of a control 20 apparatus 100 and a control target apparatus 10 according to a first embodiment. FIG. 2 is a diagram schematically showing the operation of a binary output unit 103 according to the first embodiment. FIG. 3 is a configuration diagram of a standard-deviation estimating unit 106 according to 25 the first embodiment. [0020] As shown in FIG. 1, the control target apparatus 10 includes an inter-roll conveying mechanism 11 that conveys and winds a conveyed material 12, a tension-axis-speed controller 21 that controls rotating speed of a 30 tension axis motor 13, a speed-axis-speed controller 22 that controls rotating speed of a speed axis motor 15, and an adder 23 that performs addition. A tension-axis-speed addition value Vadd is input to the control target Docket No.: PMAA-15086-PCT 8 apparatus 10 from the control apparatus 100. A reference-speed command value VrO is input to the control target apparatus 10 from the outside. [0021] The inter-roll conveying mechanism 11 includes a 5 mechanism for conveying a belt-like or linear conveyed material 12 among a plurality of rolls. As the conveyed material 12, paper, resin, metal, or fiber is illustrated. The inter-roll conveying mechanism 11 includes the tension axis motor 13 including a driving mechanism, a tension axis 10 roll 14 including a rotating mechanism, a speed axis motor 15 including a driving mechanism, a speed axis roll 16 including a rotating mechanism, and a tension detector 20 that calculates a tension detection value Yfb. [0022] The inter-roll conveying mechanism 11 drives the 15 tension axis motor 13 and the speed axis motor 15 to rotate the tension axis roll 14 and the speed axis roll 16 to wind the conveyed material 12. [0023] It is assumed that a slip between the tension axis roll 14 and the conveyed material 12 is very small. 2 0 It is assumed that circumferential speed of the tension axis roll 14 and speed of a portion of the conveyed material 12 in contact with the tension axis roll 14 coincide with each other or fit within a predetermined range. It is assumed that a slip between the speed axis 25 roll 16 and the conveyed material 12 is very small. Circumferential speed of the speed axis roll 16 and speed of a portion of the conveyed material 12 in contact with the speed axis roll 16 coincide with each other or fit within a predetermined range. 30 [0024] The tension-axis-speed controller 21 controls rotating speed of the tension axis motor 13 such that speed of conveyance of the conveyed material 12 by the tension axis roll 14 coincides with an input tension-axis-speed Docket No.: PMAA-15086-PCT 9 command value Vrl. Specifically, the tension-axis-speed controller 21 controls, taking into account the diameter or a speed reduction ratio of the tension axis roll 14, the rotating speed of the tension axis motor 13 to coincide 5 with or fit within a predetermined range with respect to a command obtained by converting the tension-axis-speed command value Vrl into the rotating speed of the tension axis motor 13. [0025] The speed-axis-speed controller 22 controls the 10 rotating speed of the speed axis motor 15 such that speed of conveyance of the conveyed material 12 by the speed axis roll 16 coincides with or fits within a predetermined range with respect to the input reference-speed command value VrO Specifically, the speed-axis-speed controller 22 controls, 15 taking into account the diameter and a speed reduction ratio of the speed axis roll 16, the rotating speed of the speed axis motor 15 to coincide with or fit within a predetermined range with respect to a command obtained by converting the reference-speed command value VrO into the 20 rotating speed of the speed axis motor 15. [0026] The reference-speed command value VrO input from the outside specifies conveying speed of the conveyed material 12 and can take various values according to conveyance conditions of the conveyed material 12. 25 [0027] The adder 23 adds up the reference-speed command value VrO and the tension-axis-speed addition value Vadd to calculate the tension-axis-speed command value Vrl and outputs the tension-axis-speed command value Vrl. [0028] The tension detector 20 outputs the tension 30 detection value Yfb, which is a value obtained by detecting the tension of the conveyed material 12. The tension detection value Yfb is a control value and is a variable controlled to approach a command value as explained below. Docket No.: PMAA-15086-PCT 10 [0029] In the first embodiment, the tension axis roll 14 is explained as a configuration for winding the conveyed material 12 and the speed axis roll 16 is explained as a configuration for unwinding the conveyed material 12. 5 However, the present invention is not limited to these configurations. The speed axis roll 16 can be configured to wind the conveyed material 12 and the tension axis roll 14 can be configured to unwind the conveyed material 12. The tension axis roll 14 and the speed axis roll 16 can be 10 intermediate shafts that do not directly perform the winding and the unwinding of the conveyed material 12 and perform only feeding operation of the conveyed material 12. [0 030] The control target apparatus 10 adds up the reference-speed command value VrO and the tension-axis- 15 speed addition value Vadd and generates the tension-axis-speed command value Vrl of the tension axis roll 14. Therefore, the tension axis roll 14 rotates faster by the tension-axis-speed addition value Vadd compared with the speed axis roll 16. That is, conveying speed of the 20 portion of the conveyed material 12 in contact with the tension axis roll 14 is higher than conveying speed of the portion of the conveyed material 12 in contact with the speed axis roll 16. The conveyed material 12 is pulled between the tension axis roll 14 and the speed axis roll 16 25 whereby tension is generated. In the control target apparatus 10, when a value of the tension-axis-speed addition value Vadd is changed, because the rotating speed of the tension axis roll 14 is changed, the tension generated in the conveyed material 12 changes. 30 [0031] The inter-roll conveying mechanism 11 measures, with the tension detector 20, tension acting on the conveyed material 12 and outputs the tension detection value Yfb. That is, the control target apparatus 10 is Docket No.: PMAA-15086-PCT 11 configured to perform feedback control by being combined with the control apparatus 100 that calculates the tension-axis-speed addition value Vadd using a value of the tension detection value Yfb. 5 [0032] The control apparatus 100 includes a control computing unit 101 that calculates an operation amount Uc, an adjustment-execution-command generating unit 102 that generates an adjustment-execution command value SWat, which is a command value indicating possibility of execution of 10 adjustment, a binary output unit 103 that calculates an adjustment-time addition value Uadd, which is a value added to the operation amount Uc during the adjustment, and a control-gain computing unit 104 that calculates a gain candidate value. The control apparatus 100 includes a 15 control-gain adjusting unit 105 that changes a value of a gain used in the control computing unit 101, a standard-deviation estimating unit 106 that calculates a standard-deviation estimated value Sig, which is an estimated value of a standard deviation, a hysteresis-width computing unit 20 107 that calculates a hysteresis-width computed value He, a subtracter 108 that performs subtraction, and an adder 109 that performs addition. A tension command value Yr is input to the control apparatus 100 from the outside. The tension detection value Yfb is input to the control 25 apparatus 100 from the control target apparatus 10. The control apparatus 100 outputs the tension-axis-speed addition value Vadd to the control target apparatus 10. [0033] A tension deviation value Ye, which is a deviation between the tension command value Yr and the 30 tension detection value Yfb, and an adjustment-execution command value SWat are input to the control computing unit 101. The tension deviation value Ye means a control deviation. In a normal state in which the adjustment- Docket No.: PMAA-15086-PCT 12 execution command value SWat is off, the control computing unit 101 outputs, as the operation amount Uc, a sum of proportional compensation obtained by multiplying the tension deviation value Ye with a proportional gain, which 5 is one of control gains, and integral compensation obtained by multiplying the tension deviation value Ye with an integral gain, which is one of the control gains, and performing integration. In the first embodiment, a period in which the adjustment-execution command value SWat is on 10 is referred to as automatic adjustment period. [0034] When the adjustment-execution command value SWat is switched from OFF to ON, the control computing unit 101 retains a value of the operation amount Uc immediately before the adjustment-execution command value SWat is 15 switched to ON and outputs the retained value of the operation amount Uc in the automatic adjustment period. Operation for retaining the value of the operation amount Uc immediately before the adjustment-execution command value SWat is switched to ON is realized by setting the 20 proportional gain and the integral gain to 0 and retaining an output of the integration. Consequently, the control computing unit 101 can keep an immediately preceding stable control state even in the automatic adjustment period, stably shift to the automatic adjustment period in which 25 automatic adjustment is executed as explained below, and set the control gain to an accurate value in a short time. [0035] The adjustment-execution-command generating unit 102 generates the adjustment-execution command value SWat, which is a signal indicating ON or OFF on the basis of an 30 instruction input by operation from the outside. The adjustment-execution-command generating unit 102 changes the adjustment-execution command value SWat from OFF to ON according to the operation from the outside, outputs a Docket No.; PMAA-15086-PCT 13 signal of ON for a predetermined period, and thereafter returns the adjustment-execution command value SWat to OFF. The predetermined period is a predetermined fixed time or a period until it is determined that an output of the binary 5 output unit 103 explained below has changed a predetermined number of times. However, the present invention is not limited to these configurations. [0036] The binary output unit 103 performs adjustment operation in the automatic adjustment period in which the 10 adjustment-execution command value SWat is ON and outputs, using the tension deviation value Ye and a hysteresis-width setting value Hs, the adjustment-time addition value Uadd, which is a value having amplitude of the magnitude of a predetermined addition value amplitude D, plus and minus of 15 the value being determined by a method explained below. Note that, as explained below, the binary output unit 103 outputs the adjustment-time addition value Uadd of 0 in a period in which the adjustment-execution command value SWat is OFF. 20 [0037] FIG. 2 is a diagram schematically showing the operation of the binary output unit 103 that selects one of +D and -D on the basis of the tension deviation value Ye and the hysteresis-width setting value Hs and outputs the adjustment-time addition value Uadd. 25 [0038] A method of determining plus and minus of the adjustment-time addition value Uadd is explained. When the magnitude of the tension deviation value Ye is larger than the hysteresis-width setting value Hs, the binary output unit 103 selects one of two values of +D and -D on the 30 basis of a sign of the tension deviation value Ye. When the magnitude of the tension deviation value Ye is equal to or smaller than the hysteresis-width setting value Hs, the binary output unit 103 selects +D if the last adjustment- Docket No.: PMAA-15086-PCT 14 time addition value Uadd is +D and selects -D if the last adjustment-time addition value Uadd is -D. That is, in the determination of the adjustment-time addition value Uadd, the tension deviation value Ye has a hysteresis 5 characteristic of the magnitude of the hysteresis-width setting value Hs. [0039] FIG. 2 schematically shows the operation of the binary output unit 103 when selecting one of the two values +D and -D on the basis of the tension deviation value Ye 10 and the hysteresis-width setting value Hs. When selecting one of +D and -D, the binary output unit 103 can select, using a signal obtained by causing a low-pass filter to act on the tension deviation value Ye instead of the tension deviation value Ye, one of the two values +D and -D on the 15 basis of the signal obtained by causing the low-pass filter to act on the tension deviation value Ye and the hysteresis-width setting value Hs. [0040] The operation of the binary output unit 103 is the same as a method called limit cycle method used in 20 temperature adjustment control. When the adjustment-execution command value SWat is switched to ON, the adjustment-time addition value Uadd output by the binary output unit 103 and the tension deviation value Ye oscillate. 25 [0041] The tension deviation value Ye and the adjustment-execution command value SWat are input to the control-gain computing unit 104. In the automatic adjustment period in which the adjustment-execution command value SWat is ON, the control-gain computing unit 104 30 measures the vibration cycle and the amplitude of the tension deviation value Ye and calculates, on the basis of a result of the measurement, a proportional-gain candidate value Gl, which is a candidate of the proportional gain of Docket No.: PMAA-15086-PCT 15 the control computing unit 101 and an integral-gain candidate value G2, which is a candidate of the integral gain. Note that the proportional gain candidate value Gl and the integral-gain candidate value G2 are collectively 5 referred to as control gain candidate value. [0042] Specifically, the control-gain computing unit 104 sets, as a corrected amplitude Ya, a value obtained by subtracting the hysteresis-width setting value Hs from the amplitude of the tension deviation value Ye and multiplies 10 the inverse of the corrected amplitude Ya with a predetermined constant to calculate the proportional-gain control value Gl. The control-gain computing unit 104 multiplies a vibration cycle of an integral time constant of proportional integral computation with a ratio of 15 amplitudes of the corrected amplitude Ya and the tension deviation value Ye and a predetermined constant to calculate the integral-gain candidate value G2. [0043] As a specific calculation method for the proportional gain and the integral gain, a method of 20 calculating a linearized gain of input and output of the binary output unit 103 on the basis of a describing function method and determining the proportional gain and the integral gain on the basis of a limit sensitivity method of Ziegler-Nichols only has to be used. According 25 to this method, the control-gain computing unit 104 is capable of performing accurate adjustment based on characteristics of the conveyed material 12 and characteristics of the tension detector 20. [0044] The control-gain computing unit 104 outputs the 30 calculated proportional-gain candidate value Gl and the calculated integral-gain candidate value G2 at a point in time when the adjustment-execution command value SWat is switched to OFF. Docket No.: PMAA-15086-PCT 16 [0045] The proportional-gain candidate value Gl and the integral-gain candidate value G2 calculated by the control-gain computing unit 104 are input to the control-gain adjusting unit 105. The control-gain adjusting unit 105 5 changes the proportional gain and the integral gain of the control computing unit 101 to the calculated proportional-gain candidate value Gl and the calculated integral-gain candidate value G2. [0046] In the first embodiment, the control-gain 10 adjusting unit 105 changes the proportional gain and the integral gain of the control computing unit 101 to the proportional-gain candidate value Gl and the integral-gain candidate value G2 immediately after the proportional-gain candidate value Gl and the integral-gain candidate value G2 15 are input. However, the present invention is not limited to these configurations. The control-gain adjusting unit 105 can execute changing operation for the proportional gain and the integral gain of the control computing unit 101 after the operator of the control apparatus 100 checks 20 the proportional-gain candidate value Gl and the integral-gain candidate value G2. The control-gain adjusting unit 105 can retain a plurality of sets of the proportional-gain candidate values Gl and the integral-gain candidate values G2 and, after the operator of the control apparatus 100 25 selects one set of the proportional gain candidate and the integral-gain candidate value G2 out of the sets, execute changing processing for the proportional gain and the integral gain of the control computing unit 101. [0047] FIG. 3 is a specific configuration diagram of the 30 standard-deviation estimating unit 106. The standard-deviation estimating unit 106 includes a high-pass filter unit 106a that calculates a low-frequency-component removed signal yh obtained by removing low-frequency components of Docket No.: PMAA-15086-PCT 17 the tension detection value Yfb, a square-value calculating unit 106b that calculates an output signal yiosb/ which is a square value of the input signal yh, and a low-pass filter unit 106c that causes a low-pass filter to act and 5 calculates an output signal yioec- The standard-deviation estimating unit 10 6 includes a square-root calculating unit 106d that calculates an output signal yioed/ which is a square root of the input signal yioec/ a selector unit 106e that outputs the standard-deviation estimated value Sig on 10 the basis of the adjustment-execution command value SWat, and a delay unit 106f that retains the standard-deviation estimated value Sig. [004 8] The adjustment-execution command value SWat and the tension detection value Yfb are input to the standard- 15 deviation estimating unit 106. In a period in which the adjustment-execution command value SWat is OFF, the standard-deviation estimating unit 106 calculates the standard-deviation estimated value Sig on the basis of the tension detection value Yfb according to a method explained 20 below and outputs the standard-deviation estimated value Sig. [0049] In the automatic adjustment period in which the adjustment-execution command value SWat is ON, the standard-deviation estimating unit 106 operates such that a 25 last standard-deviation estimated value Sig-, which is a standard-deviation estimated value in the immediately preceding cycle explained below, is output. Therefore, the standard-deviation estimated value Sig output from the standard-deviation estimating unit 106 in the automatic 30 adjustment period changes to the last standard-deviation estimated value Sig-. [0050] The tension detection value Yfb is input to the high-pass filter unit 10 6a. The high-pass filter unit 10 6a Docket No.: PMAA-15086-PCT 18 causes the high-pass filter to act on the tension detection value Yfb, calculates a low-frequency-component removed signal yh obtained by removing low-frequency components of the tension detection value Yfb, and outputs the low-5 frequency-component removed signal yh. [0051] The low-frequency-component removed signal yh is input to the square-value calculating unit 106b. The square-value calculating unit 106b calculates the output signal yioeb obtained by raising a value of the low- 10 frequency-component removed signal yh to the second power and outputs the output signal yioeb- [00 52 ] The output signal yioeb is input to the low-pass filter unit 106c. The low-pass filter unit 106c causes the low-pass filter to act on the output signal yioeb/ 15 calculates the output signal yio6c obtained by removing high-frequency components of the output signal yioeb/ and outputs the output signal yioec- [0053] The output signal yioec is input to the square-root calculating unit 106d. The square-root calculating 20 unit 106d calculates the output signal yioed obtained by calculating a square root of the output signal yioec and outputs the output signal yioed- [0054] The last standard-deviation estimated value Sig-and the output signal yioed are input to the selector unit 25 106e. The selector unit 106e selects the last standard-deviation estimated value Sig- when the adjustment-execution command value SWat is ON or selects the output signal yioed when the adjustment-execution command value SWat is OFF on the basis of the adjustment-execution 3 0 command value SWat, changes the selected output signal to the standard-deviation estimated value Sig, and outputs the standard-deviation estimated value Sig. [0055] The delay unit 106f retains the standard- Docket No.: PMM-15086-PCT 19 deviation estimated value Sig for a period of one cycle of an operation cycle of the standard-deviation estimating unit 106, changes the retained value to the last standard-deviation estimated value Sig-, and outputs the last 5 standard-deviation estimated value Sig- after one cycle of the operation cycle of the standard-deviation estimating unit 106. [0056] Referring back to FIG. 1, the standard-deviation estimated value Sig is input to the hysteresis-width 10 computing unit 107. The hysteresis-width computing unit 107 calculates the hysteresis-width computed value He on the basis of the standard-deviation estimated value Sig. Specifically, the hysteresis-width computing unit 107 calculates the hysteresis-width computed value He by 15 multiplying the standard-deviation estimated value Sig with a predetermined coefficient Kh. The coefficient Kh is a constant and is set to a value equal to or larger than 1 and equal to or smaller than 60. [0057] The hysteresis-width computing unit 107 changes 20 the hysteresis-width setting value Hs of the binary output unit 103 to the hysteresis-width computed value He at every operation cycle. [0058] The tension command value Yr and the tension detection value Yfb are input to the subtracter 108. The 2 5 subtracter 108 calculates the tension deviation value Ye from a difference between the tension command value Yr and the tension detection value Yfb, and outputs the tension deviation value Ye. [0059] The operation amount Uc and the adjustment-time 30 addition value Uadd are input to the adder 109. The adder 109 adds up the operation amount Uc and the adjustment-time addition value Uadd to calculate the tension-axis speed addition value Vadd and outputs the tension-axis speed Docket No.: PMAA-15086-PCT 20 addition value Vadd. [00 60] The operation of the control apparatus 100 is explained. In a period in which the adjustment-execution command value SWat output from the adjustment-execution-5 command generating unit 102 is OFF, the control computing unit 101 calculates the operation amount Uc to set the tension deviation value Ye to 0. The adjustment-time addition value Uadd, which is an output of the binary output unit 103 is 0. A value of the tension-axis speed 10 addition value Vadd coincides with a value of the operation amount Uc. That is, operation for outputting the tension-axis speed addition value Vadd by the control apparatus 100 is operation of feedback control by the PI control. [00 61] In the period in which the adjustment-execution 15 command value SWat is OFF, the standard-deviation estimating unit 106 outputs the standard-deviation estimated value Sig, which is an estimated value of a standard deviation of the low-frequency components removed signal yh obtained by removing low-frequency components of 2 0 the tension detection value Yfb. The hysteresis-width computing unit 107 calculates the hysteresis-width computed value He on the basis of the standard-deviation estimated value Sig calculated by the standard-deviation estimating unit 106 and changes the hysteresis-width setting value Hs 25 of the binary output unit 103 to the hysteresis-width computed value He. [0062] Operation at a point in time when the adjustment-execution-command generating unit 102 changes an output of the adjustment-execution command value SWat from OFF to ON 30 is explained. When the adjustment-execution command value SWat is changed from OFF to ON, the control computing unit 101 retains an output of integration and outputs the fixed operation amount Uc. The standard-deviation estimating Docket No.: PMAA-15086-PCT 21 unit 106 is switched to operation for substituting the last standard-deviation estimated value Sig- in the standard-deviation estimated value Sig and operates such that the last standard-deviation estimated value Sig- is output. 5 Therefore, when the adjustment-execution command value SWat is changed from OFF to ON, the standard-deviation estimated value Sig output from the standard-deviation estimating unit 106 changes to the last standard-deviation estimated value Sig-. Therefore, the hysteresis-width computed value 10 He calculated by the hysteresis-width computing unit 107 changes to a fixed value. The hysteresis-width setting value Hs of the binary output unit 103 changes to a fixed value. The binary output unit 103 alternately selects a value of +D or -D on the basis of the tension deviation 15 value Ye and the hysteresis-width setting value Hs and outputs the adjustment-time addition value Uadd. [00 63] In the selection of the value of +D or -D of the adjustment-time addition value Uadd of the binary output unit 103, when the influence of noise included in the 20 tension detection value Yfb can be appropriately removed in the hysteresis-width setting value Hs, the tension-axis-speed addition value Vadd and the tension deviation value Ye generate limit cycle vibration of a constant cycle or a cycle regarded as constant. 25 [0064] In the period in which the adjustment-execution command value SWat is ON, the control-gain computing unit 104 calculates the proportional-gain candidate value Gl and the integral-gain candidate value G2 on the basis of a vibration amplitude and a vibration cycle of the tension 30 deviation value Ye. [0065] When a predetermined automatic adjustment period elapses after the adjustment-execution command value SWat is changed from OFF to ON, the adjustment-execution-command Docket No.: PMAA-15086-PCT 22 generating unit 102 changes the adjustment-execution command value SWat from ON to OFF. [0066] At a point in time when the adjustment-execution-command generating unit 102 changes the adjustment-5 execution command value SWat from ON to OFF, the binary output unit 103 retains a value of the adjustment-time addition value Uadd at 0. The control-gain computing unit 104 outputs the proportional-gain candidate value Gl and the integral-gain candidate value G2 calculated immediately 10 before the adjustment-execution command value SWat is changed from ON to OFF. [0067] The control-gain adjusting unit 105 changes a value of the proportional gain and a value of the integral gain of the control computing unit 101 to the input 15 proportional-gain candidate value Gl and the input integral-gain candidate value G2. At this point, the standard-deviation estimating unit 106 operates such that the output signal yio6d calculated by the square-root calculating unit 106d is output. Therefore, the standard- 20 deviation estimated value Sig output from the standard-deviation estimating unit 106 is the output signal yioed-The control computing unit 101 starts calculation of the operation amount Uc on the basis of the proportional gain and the integral gain changed by the control-gain adjusting 25 unit 105. [0068] As explained below, the control apparatus 100 sets the hysteresis-width setting value Hs of the binary output unit 103 to an appropriate value and determines the adjustment-time addition value Uadd on the basis of the 30 tension deviation value Ye and the hysteresis-width setting value Hs. Therefore, the influence of noise included in the tension detection value Yfb is reduced. As a result, ■the control apparatus 100 can generate a limit cycle of a Docket No.: PMM-15086-PCT 23 constant cycle or a cycle regarded as constant and accurately adjust a control gain. [0069] The control apparatus 100 substitutes the calculated hysteresis-width computed value He in the 5 hysteresis-width setting value Hs. At this point, the hysteresis-width computed value He is calculated on the basis of the standard-deviation estimated value Sig. The standard-deviation estimated value Sig is estimated as a satisfactory estimated value of a standard deviation of 10 signals of noise included in the tension detection value Yfb by the standard-deviation estimating unit 106 explained below. That is, the control apparatus 100 estimates a value of the standard deviation of the signals of the noise included in the tension detection value Yfb on the basis of 15 the standard-deviation estimated value Sig. Therefore, it is possible to estimate a distribution of the signals of noise that occurs in the automatic adjustment period. It is possible to calculate the hysteresis-width computed value He having magnitude of a degree larger than the 20 amplitude of the signals of the noise at a high probability [0070] The operation and the effect of the standard-deviation estimating unit 106 are explained. In the period in which the adjustment-execution command value SWat is OFF, the high-pass filter unit 106a calculates the low- 25 frequency-component removed signal yh obtained by causing the high-pass filter to act on the tension detection value Yfb. [0071] At this point, if a control gain of the control computing unit 101 is not an appropriate value and a 30 control band is low, the tension detection value Yfb has a vibration component of a low frequency due to the influence of disturbance and an offset error with respect to the tension command value Yr. The vibration component of the Docket No.: PMM-15086-PCT 24 low frequency and the offset error occur because the control band is low and control performance is inappropriate and are phenomena not seen during operation of the limit cycle method for vibrating the tension 5 detection value Yfb at a high frequency near a limit frequency. That is, the vibration component of the low frequency and the offset error caused because the control band is low do not affect determination of the adjustment-time addition value Uadd in the limit cycle method. 10 Therefore, it is unnecessary to taken into account the vibration component of the low frequency and the offset error in the calculation of the hysteresis-width computing value He. Further, the amplitude of the vibration of the low-frequency components of the tension detection value Yfb 15 is sometimes large compared with the amplitude of the noise If the standard-deviation estimated value Sig is calculated without distinguishing the vibration of the low-frequency components of the tension detection value Yfb and the signals of the noise, a calculation result of the standard- 20 deviation estimated value Sig is considerably larger than a true value of a standard deviation of the signals of the noise. The hysteresis-width setting value Hs larger more than necessary is set and adversely affects the vibration amplitude of the tension detection value Yfb to be 25 excessively amplified. Therefore, the vibration component of the low-frequency components and the offset error should be removed when the standard-deviation estimated value Sig is calculated. [0072] The high-pass filter unit 106a causes the high- 30 pass filter to act on the tension detection value Yfb. Therefore, it is possible to remove the vibration component of the low frequency and the offset error of the tension detection value Yfb and extracts, with the low-frequency- Docket No.: PMAA-15086-PCT 25 component removed signal yh, the signals of the noise consisting of the high-frequency components of the tension detection value Yfb. [0073] The configuration and the effect of the standard-5 deviation estimating unit 106 obtained by including the square-value calculating unit 106b, the low-pass filter unit 106c, and the square-root calculating unit 106d are explained. For the explanation, time series signals x of a certain normal distribution are explained as an example. 10 An average of the time series signals x is represented as fj.x and a standard deviation of the time series signals x is represented as crx. Because samples are extracted from the time series signals x in time series order, when a time interval of the samples is represented as dt, the following 15 Expression (1) and Expression (2) hold: [0074] [Math. 1] 1 w-i u,x = lim — X xs (n - k) x dt (1) [0075] 20 [Math. 2] \ 1 w-i ~ ~ a = Jlim — X xs(n — k)1 x dt — fi ( 2 ) V A'-*M T *=o [0076] In Expression (1) and Expression (2), xs(i) indicates an i-th sample extracted from the time series signals x and xs(n) indicates a sample at a terminal end 25 time in the time series signals x. In Expression fl) and Expression (2), T-Nxdt. [0077] On the other hand, an i-th sample of a signal obtained by causing a first-order lag low-pass filter, which has T larger than dt as a time constant, to act on 30 the time series signals x is represented as xlpf(i). At this point, xlpf(n) at the terminal end time is calculated 26 Docket No.: PMAA-15086-PCT by the following Expressions (3). [0078] [Math. 3] xlpf(n) = — x xs(n) + 1 x xlpf(n - 1) (3) 5 [0079] The number of samples of the time series signals x used for the calculation of xlpf(p) is represented as N. When N is large, the following Expression (4) holds. [0080] [Math. 4] xlpf(n) - "l— x fl - —) x xs{n ~ k) + (l - —] x xlpf(n - N) *=0 X V X J V X ) 2dt ]_ W-] = — X ■^■s(n - ic) x cit + o x *■=(> (4) [0081] The expression is approximated making use of the fact that the time constant x is larger than dt and N is larger than 1. In the expression, o(dt2/x2) is obtained by 15 putting in order terms having sizes equal to or smaller than the order of dt2/t2 and can be approximated to 0. [0082] When Expression (1) and Expression (4) are compared, when the number N of samples is large and the time constant x of the low-pass filter is larger than the 20 time interval dt of the samples, xlpf (n) and fj.x coincide with each other or fit within a predetermined range, it is seen that xlpf(n) is a good estimated value of \xx. [0083] Similarly, when a result obtained by causing the first-order lag low frequency filter, which has x as the 2 5 time constant, to act on a value obtained by raising the sample xs(n) to the second power is represented as x21pf(n) x21pf(n) is calculated by the following Expression (5). [0084] [Math. 5] Docket No.: PMAA-15086-PCT 27 x21pf(n) = — x xs(nf + (l ~ — | x x21pf(n ~ 1) T V X 'dt^ 5 1 w-l , lira — Z xs(n - ky x dt + o A<->« T ^ = 0 [0085] What is represented by A in Expression (6-1) is defined by Expression (6-2) using x21pf(n) and xlpf(n). [00861 6-1 [Math. 6] A = ax(n) [0087] [Math. 7] 10 15 ax(n) = sjx21pf(n) - xlpf(nf (6-2) [0088] When Expression (2) and Expression (6-2) are compared, when the number N of samples is large and the time constant t of the low-pass filter is larger than the time interval dt of the samples, the following fact is found. [0089] That is, because A and cx coincide with each other or fit within a predetermined range, it is seen that A is a good estimated value of ax. If the average u.x of the time series signals x is 0, it is evident that Expression (7) is obtained. [0090] [Math. 8] 25 30 ax(n) - [0148] Therefore, a value obtained by multiplying the standard deviation oax of the signals ax with V (re/ (TT-2 } ) is a standard deviation of the time series signals x. Note that V (n/ (TT-2 ) ) means that [n/(n-2) ) is a radical sign. 25 [0149] Samples are extracted from the signals ax at the time interval dt. An i-th sample is represented as axs(i). A sample of the signals ax at the terminal end time is represented as axs (n) . An i-th sample of a signal obtained by causing a first-order lag low-pass filter, which has T 30 larger than dt as a time constant, to act on the signals ax is represented as axlpf(i). The standard deviation aax of Docket No.: PMM-15086-PCT 45 the signals ax is calculated by the following Expression (14) . [0150] [Math. 15] I 1 «-i , ~™ „ 5 Gax = Jlim ™ ^axsin - Jc)' x dt - \xa^; (14) V N-"n T if=0 where, T=Nxdt. [0151] When a result obtained by causing the first-order lag low frequency filter, which has x as the time constant, to act on a value obtained by raising the sample axs(n) to 10 the second power is represented as ax21pf(n), ax21pf(n) is calculated by the following Expression (15). [0152] [Math. 16] ax21pf(n) - lim- "faxsfn - k)2 x dt (15) w->« x k=0 15 [0153] What is represented by B in Expression (16-1) is defined by Expression (16-2) using axlpf(n) and ax21pf(n) calculated by Expression (12) and Expression (15). [0154] [Math. 17] 20 B = cax(n) (16-1) [0155] [Math. 18] oax(n) = Jax21pf(n) - axlpf(nf (16-2) [0156] When Expression (14) and Expression (16-2) are 25 compared, it is seen that, because B and aax coincide with each other or fit within a predetermined range when the number N of samples is large and the time constant T of the low-pass filter is larger than the time interval dt of the samples, B is a good estimated value of aax. 30 [0157] The absolute-value computing unit 116b, the first low-pass filter unit 126c, the first square-value Docket No.: PMAA-15086-PCT 46 calculating unit 126d, the second square-value calculating unit 12 6e, the second low-pass filter unit 126f, the subtracter 126g, the square-root calculating unit 126h, and the conversion gain device 126i perform calculations same 5 as the calculations in the example explained above. That is, the output signal yi26h is a good estimated value of the standard deviation of the output signal yneb- The output signal yi26i is a good estimated value of the standard deviation of the low-frequency-component removed signal yh. 10 [0158] The control apparatus 300 can set hysteresis having appropriate magnitude and appropriately determines an adjustment-time addition value in the automatic adjustment period. Therefore, it is possible to suppress hunting due to the influence of the noise included in the 15 tension detection value Yfb. As a result, the control apparatus 300 can generate a limit cycle of a constant cycle or a cycle regarded as constant and accurately calculate a value of a control gain. [0159] Fourth Embodiment. 20 A fourth embodiment of the control apparatus according to the present invention is explained. FIG. 8 is a configuration diagram of a control apparatus 400 and the control target apparatus 10 according to the fourth embodiment. FIG. 9 is a configuration diagram of a 25 standard-deviation estimating unit 136. The standard-deviation estimating unit 136 is obtained by adding a change to the standard-deviation estimating unit 106 of the control apparatus 100. Note that components having functions same as the functions in the first embodiment are 30 denoted by reference numerals and signs same as the reference numerals and signs in the first embodiment or the second embodiment and detailed explanation of the components is omitted. Docket No.: PMAA-15086-PCT 47 [0160] The control apparatus 4 00 includes the control computing unit 101 that calculates the operation amount 0c, the adjustment-execution-command generating unit 102 that generates the adjustment-execution command value SWat, the 5 binary output unit 103 that calculates the adjustment-time addition value Uadd, and the control-gain computing unit 104 that calculates a gain candidate value. The control apparatus 400 includes the control-gain adjusting unit 105 that changes a value of a gain used in the control 10 computing unit 101, the standard-deviation estimating unit 136 that calculates the standard-deviation estimated value Sig, the hysteresis-width computing unit 107 that calculates the hysteresis-width computed value He, the subtracter 108 that performs subtraction, and the adder 109 15 that performs addition. [0161] The tension command value Yr is input to the control apparatus 400 from the outside. The tension detection value Yfb is input to the control apparatus 4 00 from the control target apparatus 10. The control 2 0 apparatus 4 00 outputs the tension-axis-speed addition value Vadd to the control target apparatus 10. [0162] The standard-deviation estimating unit 136 includes the high-pass filter unit 106a that calculates the low-frequency-component removed signal yh, a data retaining 25 unit 13 6b that retains the low-frequency-component removed signal yh, a standard-deviation computing unit 136c that calculates a standard deviation, a selector unit 136d that outputs the standard-deviation estimated value Sig on the basis of the adjustment-execution command value SWat, and 30 the delay unit 106f that retains the standard-deviation estimated value Sig. [0163] The adjustment-execution command value SWat and the tension detection value Yfb are input to the standard- Docket No.: PMM-15086-PCT 48 deviation estimating unit 136. In the period in which the adjustment-execution command value SWat is OFF, the standard-deviation estimating unit 136 calculates the standard-deviation estimated value Sig on the basis of the 5 tension detection value Yfb and outputs the standard-deviation estimated value Sig. [0164] In the automatic adjustment period in which the adjustment-execution command value SWat is ON, the standard-deviation estimating unit 136 operates such that 10 the last standard-deviation estimated value Sig- is output. Therefore, the standard-deviation estimated value Sig output from the standard-deviation estimating unit 136 in the automatic adjustment period changes to the last standard-deviation estimated value Sig-. 15 [0165] The low-frequency-component removed signal yh is input to the data retaining unit 136b. The data retaining unit 13 6b retains a value of the low-frequency-component removed signal yh as data associated with time, changes latest m data among the retained data to an output signal 20 yi36b/ and outputs the output signal yi36b- Note that m is an integer equal to or larger than 1, [0166] The output signal yi36b is input to the standard-deviation computing unit 136c. The standard-deviation computing unit 136c calculates a standard deviation of the 25 output signal yi36b an^ outputs a calculated output signal yi36c- The output signal yi36b is m low-frequency-component removed signals yh associated with time and is the low-frequency-component removed signal yh in a predetermined period. Therefore, in the following explanation, the 30 standard deviation of the output signal yi^eb is referred to as a standard deviation in the predetermined period. [0167] The last standard-deviation estimated value Sig-and the output signal yi36c are input to the selector unit 49 Docket No.: PMAA-15086-PCT 136d. On the basis of the adjustment-execution command value SWat, the selector unit 136d selects the last standard-deviation estimated value Sig- when the adjustment-execution command value SWat is ON and selects 5 the output signal yi36c when the adjustment-execution command value SWat is OFF, changes the selected value to the standard-deviation estimated value Sig, and outputs the standard-deviation estimated value Sig. [0168] That is, in the period in which the adjustment- 10 execution command value SWat is OFF, the standard-deviation estimating unit 13 6 outputs the standard-deviation estimated value Sig, which is the estimated value of the standard deviation of the low-frequency-component removed signal yh obtained by removing the low-frequency components 15 of the tension detection value Yfb. [0169] The control apparatus 400 sets the hysteresis-width setting value Hs of the binary output unit 103 to an appropriate value and determines the adjustment-time addition value Uadd on the basis of the tension deviation 20 value Ye and the hysteresis-width setting value Hs. Therefore, the influence of the noise included in the tension detection value Yfb is reduced. As a result, the control apparatus 400 can generate a limit cycle of a constant cycle or a cycle regarded as constant and 25 accurately adjust a control gain. [0170] The control apparatus 400 substitutes the calculated hysteresis-width computed value He in the hysteresis-width setting value Hs. At this point, the hysteresis-width computed value He is calculated on the 30 basis of the standard-deviation estimated value Sig. The standard-deviation estimated value Sig is estimated as a satisfactory estimated value of a standard deviation of signals of noise included in the tension detection value Docket No.: PMAA-15086-PCT 50 Yfb by the standard-deviation estimating unit 13 6 explained below. That is, the control apparatus 400 estimates a value of the standard deviation of the signals of the noise included in the tension detection value Yfb on the basis of 5 the standard-deviation estimated value Sig. Therefore, it is possible to estimate a distribution of the signals of the noise that occurs in the automatic adjustment period. It is possible to calculate the hysteresis-width computed. value He having magnitude of a degree larger than the 10 amplitude of the signals of the noise at a high probability [0171] The control apparatus 400 and the control apparatus 100 are different only in the configuration of the standard-deviation estimating unit. Operation and effect in calculating the standard-deviation estimated 15 value Sig with the standard-deviation estimating unit 136 are explained below. [0172] As in the first, second, and third embodiments, the time series signals x of a certain normal distribution are explained as an example. A standard deviation of the 20 time series signals x is represented as ax. According to Expression (2), the standard deviation CTX is calculated when the number of samples xs is infinitely increased. Here, LIXM and axM are defined by Expression (17) and Expression (18) using a certain number M. M is an integer 25 equal to or larger than 1. [0173] [Math. 19] (ivl. - - "lx5(n - k) x dt (17) T *-o [0174] 30 [Math. 20] v,<- = J- "fxs(n - kf x dt - uw/ (18) VT ^o Docket No.: PMAA-15086-PCT 51 [0175] When Expression (2) and Expression (18) are compared, if M is a large number, axM calculated by M samples xs and the standard deviation ax coincide with each 5 other or fit within a predetermined range. Therefore, ayM is considered to be a good estimated value of the standard deviation ax. Because axM is a standard deviation of the time series signals x in a predetermined period, axM is the specified period standard deviation explained above. 10 [0176] When the number m of data of the low-frequency removed signal yh retained in the data retaining unit 136b is set to a large value, the standard-deviation computing unit 136c performs calculation same as the calculation in the example explained above. That is, the output signal 15 yi36c is considered to be a good estimated value of the standard deviation of the low-frequency-component removed signal yh. [0177] Therefore, the control apparatus 400 can set hysteresis having appropriate magnitude and appropriately 20 determines an adjustment-time addition value in the automatic adjustment period. Therefore, it is possible to suppress hunting due to the influence of the noise included in the tension detection value Yfb. As a result, the control apparatus 400 can generate a limit cycle of a 25 constant cycle or a cycle regarded as constant and accurately calculate a value of a control gain. [0178] Fifth Embodiment. A fifth embodiment of the control apparatus according to the present invention is explained. FIG. 10 is a 30 configuration diagram of a control apparatus 500 and the control'target apparatus 10 according to the fifth embodiment. FIG. 11 is a configuration diagram of a Docket No.: PMAA-15086-PCT 52 standard-deviation estimating unit 146. The standard-deviation estimating unit 146 is obtained by adding a change to the standard-deviation estimating unit 106 of the control apparatus 100. Note that components having 5 functions same as the functions in the first embodiment are denoted by reference numerals and signs same as the reference numerals and signs in the first embodiment and detailed explanation of the components is omitted. [0179] The control apparatus 500 includes the control 10 computing unit 101 that calculates the operation amount Uc, the adjustment-execution-command generating unit 102 that generates the adjustment-execution command value SWat, the binary output unit 103 that calculates the adjustment-time addition value Uadd, and the control-gain computing unit 15 104 that calculates a gain candidate value. The control apparatus 500 includes the control-gain adjusting unit 105 that changes a value of a gain used in the control computing unit 101, the standard-deviation estimating unit 14 6 that calculates the standard-deviation estimated value 20 Sig, the hysteresis-width computing unit 107 that calculates the hysteresis-width computed value He, the subtracter 108 that performs subtraction, and the adder 109 that performs addition. The tension command value Yr is input to the control apparatus 500 from the outside. The 25 tension detection value Yfb is input to the control apparatus 500 from the control target apparatus 10. The control apparatus 500 outputs the tension-axis-speed addition value Vadd to the control target apparatus 10. [0180] The standard-deviation estimating unit 14 6 30 includes a high-pass filter unit 146a that calculates the low-frequency-component removed signal yh, the square-value calculating unit 106b that calculates the output signal yioeb with square value calculation, and the low-pass filter Docket No.: PMAA-15086-PCT 53 unit 106c that causes the low-pass filter to act and calculates the output signal yioec- The standard-deviation estimating unit 146 includes the square-root calculating unit 106d that calculates the output signal yioea with 5 square root calculation, the selector unit 106e that outputs the standard-deviation estimated value Sig on the basis of the adjustment-execution command value SWat, and the delay unit 106f that retains the standard-deviation estimated value Sig. 10 [0181] The adjustment-execution command value SWat and the tension deviation value Ye are input to the standard-deviation estimating unit 146. In the period in which the adjustment-execution command value SWat is OFF, the standard-deviation estimating unit 146 calculates the 15 standard-deviation estimated value Sig on the basis of the tension deviation value Ye and outputs the standard-deviation estimated value Sig. [0182] In the automatic adjustment period in which the adjustment-execution command value SWat is ON, the 20 standard-deviation estimating unit 146 operates such that the last standard-deviation estimated value Sig- is output. Therefore, the standard-deviation estimated value Sig output from the standard-deviation estimating unit 146 in the automatic adjustment period changes to the last 25 standard-deviation estimated value Sig-. [0183] The tension deviation value Ye is input to the high-pass filter unit 146a. The high-pass filter unit 146a causes the high-pass filter to act on the tension detection value Ye, calculates the low-frequency-component removed 30 signal yh obtained by removing the low-frequency components of the tension deviation value Ye, and outputs the low-frequency-component removed signal yh. [0184] That is, in the period in which the adjustment- Docket No.: PMAA-15086-PCT 54 execution command value SWat is OFF, the standard-deviation estimating unit 146 outputs the standard-deviation estimated value Sig, which is the estimated value of the standard deviation of the low-frequency-component removed 5 signal yh obtained by removing the low-frequency components of the tension deviation value Ye. [0185] The control apparatus 500 sets the hysteresis-width setting value Hs of the binary output unit 103 to an appropriate value and determines the adjustment-time 10 addition value Uadd on the basis of the tension deviation value Ye and the hysteresis-width setting value Hs. Therefore, the influence of the noise included in the tension detection value Yfb is reduced. As a result, the control apparatus 500 can generate a limit cycle of a 15 constant cycle or a cycle regarded as constant and accurately adjust a control gain. [0186] The control apparatus 500 substitutes the calculated hysteresis-width computed value He in the hysteresis-width setting value Hs. At this point, the 20 hysteresis-width computed value He is calculated on the basis of the standard-deviation estimated value Sig. The standard-deviation estimated value Sig is estimated as a satisfactory estimated value of a standard deviation of signals of noise included in the tension detection value 25 Yfb by the standard-deviation estimating unit 14 6 explained below. That is, the control apparatus 500 estimates a value of the standard deviation of the signals of the noise included in the tension detection value Yfb on the basis of the standard-deviation estimated value Sig. Therefore, it 30 is possible to estimate a distribution of the signals of the noise that occurs in the automatic adjustment period. It is possible to calculate the hysteresis-width computed value He having magnitude of a degree larger than the Docket No.: PMAA-15086-PCT 55 amplitude of the signals of the noise at a high probability [0187] The control apparatus 500 and the control apparatus 100 are different only in the configuration of the standard-deviation estimating unit. Operation and 5 effect in calculating the standard-deviation estimated value Sig with the standard-deviation estimating unit 14 6 are explained below. [0188] The tension deviation value Ye input to the standard-deviation estimating unit 146 is obtained by 10 subtracting the tension command value Yr, which is a fixed value, from the tension detection value Yfb. Only low-frequency components are different from low™frequency components of the tension detection value Yfb. Therefore, a signal obtained by removing the low-frequency components 15 of the tension deviation value Ye coincides with or fits within a predetermined range with respect to a signal obtained by removing the low-frequency components from the tension detection value Yfb. [0189] Therefore, the standard-deviation estimating unit 20 146 is the same as the standard-deviation estimating unit 106 of the control apparatus 100 except the high-pass filter unit 146a that calculates the low-freguency-component removed signal yh. Therefore, an effect same as the effect of the standard-deviation estimating unit 106 25 can be obtained. [0190] Therefore, the control apparatus 5 00 can set hysteresis having appropriate magnitude and appropriately determines an adjustment-time addition value in the automatic adjustment period. Therefore, it is possible to 30 suppress hunting due to the influence of the noise included in the tension detection value Yfb. As a result, the control apparatus 500 can generate a limit cycle of a constant cycle or a cycle regarded as constant and Docket No.: PMAA-15086-PCT 56 accurately calculate a value of a control gain. [0191] Sixth Embodiment. FIG. 12 is a configuration diagram of a control apparatus 60 0 and a control target apparatus 3 0 according 5 to a sixth embodiment. Note that, among components of the control apparatus 600, components having functions same as the functions of the control apparatus 100 according to the first embodiment are denoted by reference numerals and signs same as the reference numerals and signs in the first 10 embodiment and detailed explanation of the components is omitted. [0192] As shown in FIG. 12, the control target apparatus 30 is a liquid heating apparatus including a water tank 31 that stores liquid 35, a temperature detector 32 that 15 detects the temperature of the liquid 35, a heater 33 that heats the liquid 35, and a current supplier 34 that supplies an electric current. The liquid heating apparatus adjusts the temperature of the liquid 35. [0193] A current command value Ir is input to the 20 control target apparatus 30 from the control apparatus 600. The control target apparatus 3 0 outputs a temperature detection value Tfb. [0194] The temperature detector 32 detects the temperature of the liquid 35 and outputs the temperature 25 detection value Tfb, which is a detected value. The temperature detection value Tfb is a control amount and is a value controlled to approach a command value as explained below. [0195] An electric current is supplied to the heater 33 30 from the current supplier 34. The heater 33 generates heat on the basis of the magnitude of the supplied electric current. The heat generated by the heater 33 is transmitted to the liquid 35 and heats the liquid 35. Docket No.: PMAA-15086-PCT 57 [0196] The current supplier 34 supplies, on the basis of the current command value Ir, an amount of an electric current coinciding with the current command value Ir to the heater 33. 5 [0197] The liquid 35 is a liquid such as water, oil, or a drug solution. Because the heat is transmitted from the heater 33, the temperature of the liquid 35 rises. That is when a value of the current command value Ir is changed, the temperature of the liquid 35 changes. 10 [0198] The control target apparatus 30 measures the temperature of the liquid 35 with the temperature detector 32 and outputs the temperature detection value Tfb. That is, the control target apparatus 30 is configured to perform feedback, control by being combined with the control 15 apparatus 600 that calculates the current command value Ir using the value of the temperature detection value Tfb. [0199] The control apparatus 600 includes a control computing unit 601 that calculates the operation amount Uc, the adjustment-execution-command generating unit 102 that 2 0 generates the adjustment-execution command value SWat, which is a command value indicating possibility of execution of adjustment, a binary output unit 603 that calculates the adjustment-time addition value Uadd, which is a value added to the operation amount Uc during the 2 5 adjustment, and a control-gain computing unit 60 4 that calculates a gain candidate value. The control apparatus 600 includes the control-gain adjusting unit 105 that changes a value of a gain used in the control computing unit 601, a standard-deviation estimating unit 606 that 30 calculates the standard-deviation estimated value Sig, the hysteresis-width computing unit 107 that calculates the hysteresis-width computed value He, a subtracter 608 that performs subtraction, and an adder 609 that performs Docket No.: PMAA-15086-PCT 58 addition. A temperature command value Tr is input to the control apparatus 600 from the outside. A temperature detection value Tfb is input to the control apparatus 600 from the control target apparatus 30. The control 5 apparatus 60 0 outputs the current command value Ir to the control target apparatus 30. [0200] The control computing unit 601, the binary output unit 603, and the control-gain computing unit 60 4 have functions same as the functions of the control computing 10 unit 101, the binary output unit 103, and the control-gain computing unit 104 in the first embodiment except that one of input signals is a temperature deviation value Te. Therefore, detailed explanation of operation is omitted. The standard-deviation estimating unit 606 has a function 15 same as the function of the standard-deviation estimating unit 106 in the first embodiment except that one of input signals is the temperature deviation value Te and the other of the input signals is the temperature detection value Tfb Therefore, detailed explanation of operation is omitted. 20 [0201] The temperature command value Tr and the temperature detection value Tfb are input to the subtracter 60 8. The subtracter 608 calculates the temperature deviation value Te from a difference between the temperature command value Tr and the temperature detection 25 value Tfb and outputs the temperature deviation value Te. [0202] The operation amount Uc and the adjustment-time addition value Uadd are input to the adder 609. The adder 609 adds up the operation amount Uc and the adjustment-time addition value Uadd to calculate the current command value 30 Ir and outputs the current command value Ir. [0203] When the adjustment execution command SWat is OFF, the control target apparatus 30 is feedback-controlled such that the temperature detection value Tfb approaches the Docket No.: PMAA-15086-PCT 59 temperature command value Tr according to the operation amount Uc calculated by the control computing unit 601. When the adjustment execution command SWat is ON and the hysteresis-width setting value Hs is appropriately set, the 5 current command value Ir and the temperature deviation value Te generate limit cycle vibration of a constant cycle or a cycle regarded as constant. [0204] The control apparatus 600 according to the sixth embodiment has a configuration same as the configuration of 10 the control apparatus 100 according to the first embodiment Therefore, the control apparatus 600 achieves an effect same as the effect achieved by the control apparatus 100. [0205] Therefore, the control apparatus 600 can set hysteresis having appropriate magnitude and appropriately 15 determines the adjustment-time addition value Uadd. Therefore, it is possible to suppress hunting due to the influence of noise included in the temperature detection value Tfb. As a result, the control apparatus 600 can generate a limit cycle of a constant cycle or a cycle 20 regarded as constant and accurately calculate a value of a control gain. 9S Reference Signs List [0206] 10, 30 control target apparatus 11 inter-roll conveying mechanism 12 conveyed material 13 tension axis motor 14 tension axis roll 15 speed axis motor 16 speed axis roll 20 tension detector 21 tension-axis-speed controller 2 2 speed-axis-speed controller Docket No.: PMAA-15086-PCT 60 23, 109, 609 adder 31 water tank 32 temperature detector 33 heater 5 34 current supplier 35 liquid 100, 200, 300, 400, 500, 600 control apparatus 101, 601 control computing unit 102 adjustment-execution-command generating unit 10 103, 603 binary output unit 104, 604 control-gain computing unit 105 control-gain adjusting unit 106, 116, 126, 136, 146, 606 standard-deviation estimating unit 15 106a, 146a high-pass filter unit 106b square-value calculating unit 106c, 116c low-pass filter unit 10 6d, 126h square-root calculating unit 106e, 116e, 126j, 136d selector unit 20 106f delay unit 107 hysteresis-width computing unit 108, 126g, 608 subtracter 116b absolute-value calculating unit 116d, 126i conversion gain device 25 126d first square-value calculating unit 126e second square-value calculating unit 126c first low-pass filter unit 126f second low-pass filter unit 136b data retaining unit 3 0 13 6c standard-deviation computing unit Docket No.: PMAA-15086-PCT 61 CLAIMS 1. A control apparatus comprising: a subtracter that calculates a control deviation on the basis of a command value input from an outside and a 5 control amount input from a control target apparatus; a control computing unit that generates an operation amount on the basis of the control deviation and a control gain and outputs the operation amount; an adjustment-execution-command generating unit that 10 generates an adjustment-execution command value indicating ON or OFF and outputs the adjustment-execution command value; a binary output unit that generates, in a period in which the adjustment-execution command value output from 15 the adjustment-execution-command generating unit is ON, an adjustment-time addition value on the basis of the control deviation and a hysteresis-width setting value and outputs the adjustment-time addition value; a standard-deviation estimating unit that calculates, 20 in a period in which the adjustment-execution command value output from the adjustment-execution-command generating unit is OFF, a low-frequency-component removed signal obtained by removing low-frequency components of the control amount or the control deviation and calculates a 25 standard-deviation estimated value, which is an estimated value of a standard deviation of the low-frequency component removed signal; and a hysteresis-width computing unit that calculates a hysteresis-width computed value on the basis of the 30 standard-deviation estimated value and changes the hysteresis-width setting value of the binary output unit to the hysteresis-width computed value. Docket No.: PMAA-15086-PCT 62 2. The control apparatus according to claim 1, wherein the standard-deviation estimating unit extracts and retains a value of the low-frequency-component removed signal in a predetermined period, calculates a standard deviation of 5 the extracted and retained value, and sets the calculated standard deviation of the predetermined period as the standard-deviation estimated value. 3. The control apparatus according to claim 1, wherein 10 the standard-deviation estimating unit calculates a signal obtained by raising the low-frequency-component removed signal to a second power, causes a low-pass filter to act on the calculated signal and calculates a signal obtained by removing high-frequency components, and calculates a 15 square root of the calculated signal to thereby calculate the standard-deviation estimated value. 4. The control apparatus according to claim 1, wherein the hysteresis-width computing unit calculates the 20 hysteresis-width computed value by multiplying the standard-deviation estimated value with a predetermined coefficient. 5. The control apparatus according to claim 1, wherein 25 the standard-deviation estimating unit causes a high-pass filter to act on the control amount or the control deviation to thereby calculate the low-frequency-component removed signal. 30 6. The control apparatus according to claim 1, further comprising: a control-gain computing unit that calculates, in the period in which the adjustment-execution command value Docket No.: PMAA-15086-PCT 63 output from the adjustment-execution-command generating unit is ON, a control-gain candidate value, which is a candidate value of the control gain, on the basis of the control deviation; and 5 a control-gain adjusting unit that changes the control gain used for the calculation of the operation amount by the control computing unit to the control-gain candidate value calculated by the control-gain computing unit. 10 7. A control method comprising: a subtracting step of performing subtraction of a command value input from an outside and a control amount input from a control target apparatus to thereby calculate a control deviation; 15 a control computing step of generating an operation amount on the basis of the control deviation and a control gain and outputting the operation amount; an adjustment-execution-command generating step of generating an adjustment-execution command value indicating 20 ON or OFF and outputting the adjustment execution command value; a binary output step of generating, in a period in which the adjustment-execution command value output from the adjustment-execution-command generating step is ON, an 25 adjustment-time addition value on the basis of the control deviation and a hysteresis-width setting value and outputting the adjustment-time addition value; a standard-deviation estimating step of calculating, in a period in which the adjustment-execution command value 30 output from the adjustment-execution-command generating step is OFF, a low-frequency-component removed signal obtained by removing low-frequency components of the control amount or the control deviation and calculating a Docket No.: PMAA-15086-PCT 64 standard-deviation estimated value, which is an estimated value of a standard deviation of the low-frequency component removed signal; and a hysteresis-width computing step of calculating a 5 hysteresis-width computed value on the basis of the standard-deviation estimated value and changing the hysteresis-width setting value of the binary output step to the hysteresis-width computed value. 10