Claims:WE CLAIM:
1. A method of real time thermal error compensation for CNC machine; said method comprising steps of-
a) calibrating the CNC machine to assess the thermal errors as a function of thermo-mechanical history, internal residual heat generation and ambient temperature; said calibration comprising-
i) fixing a metrological fixture comprising plurality of displacement sensors on table of 3-axis vertical machining center of the machine and plurality of thermal sensors onto the CNC machine;
ii) calibrating the thermal error as a function of thermo-mechanicalhistory effects by maintaining the residual/internal heat effects and ambient temperature constant;
iii) calibrating the thermal error as a function of residual heat effects by maintaining the thermo-mechanicalhistory effects and ambient temperature constant; and
iv) calibrating the thermal error as a function of ambient temperature by idling the machine;
b) communicating calibration data to controller of the CNC machine through Internet of Things software to set baseline for thermal distortion and spindle offset;
c) measuring the ambient temperature and rear bearing temperature of the spindle via PLC module during real time machining process periodically and communicating the measurement to controller of the CNC machine; and
d) updating, summing the errors during real time machining and communicating to the CNC controller using Internet of Things software to compensate for the thermal error.

2. The method according to claim 1, wherein the CNC machine is selected from a group comprising vertical machining centre, horizontal machining centre, double column machining centre, turning machines, grinding machines and special purpose machine.
3. The method according to claim 1, wherein the calibration of the thermal error as a function of thermo-mechanicalhistory effects, comprises steps of:
c) measuring ambient temperature at predefined time slots at heated and cold conditions as well as the Tool Center Point (TCP) distortion using metrological fixture periodically and communicating the data to the controller; and
d) preparing the thermo-mechanical baseline module by transformation of displacement recorded bythe metrologicalfixture to obtain thermal distortion, ?x, ?y,?z, ?R, pitch and yaw angles by the controller.
4. The method according to claim 3, wherein the error as a function of thermo-mechanical effect is considered when the ratio of temperature between rear bearing of the spindle and ambient temperature is above threshold maximum of 1.15or below threshold minimum of 0.85.
5. The method according to claim 1, wherein the calibration of the thermal error as a function of residual heat effects, comprises steps of:
a) warming up the machine and running it alternatively at constant speed and random speed;
b) measuring ambient temperature and TCP distortion at predefined time slots using metrological fixture periodically and communicating the data to the controller; and
c) preparing the residual heat baseline module by transformation of the displacement recorded by metrologicalfixture to obtain thermal distortion, ?x, ?y,?z, ?R, pitch and yaw angles by the controller.
6. The method according to claim 1, wherein the calibration of the thermal error as a function of ambient temperature, comprises steps of:
a) keeping the machine at idle state;
b) measuring ambient temperature and TCP distortion using metrological fixture periodically and communicating data to the controller; and
c) preparing the ambient temperature baseline module by transformation of the displacement recorded by metrology fixture to obtain thermal distortion, ?x, ?y,?z, ?R, pitch and yaw angles by the controller.
7. A system for compensating real time thermal error in a CNC machine during machining, said system comprising-
metrological fixture with sensors to assess displacement; thermal sensors to asses distortion due to thermo-mechanical history, internal residual heat generation and ambient temperature; Internet of Things platform communicating the thermal changes to controller of CNC machine and compensating error due to thermal changes during machining.
8. A method of developing a workpiece by a CNC machine compensating the thermal error, said method comprising-
c) measuring the ambient temperature and rear bearing temperature of the spindle via PLC module during real time machining process periodically and communicating the measurement to controller of the CNC machine; and
d) updating, summing the errors during real time machining and communicating to the CNC controller using Internet of Things software to compensate for the thermal error.
, Description:TECHNICAL FIELD
The present disclosure in general relates to thermal error compensation method for CNC machine. Particularly, the present invention provides a process for compensation of thermal error by decomposing the thermal error sources and building compensation models separately and accordingly. More particularly, the disclosure provides a method for building robust compensation model for origin shifting technique through work and tool offsets by separately considering the effects of thermo-mechanical history, ambient temperature variations and residual heat generated during machining operation.

BACKGROUND AND PRIOR ART
One of the applications of Computer Numerically Controlled (CNC) machine tool is to remove material from the work piece in order to achieve desired form and functional requirements. Although the complete subtractive machining process is CNC-controlled, one may typically encounter variations in the machined dimensions in the form of reduced accuracy and repeatability. For instance, if one considers the mass manufacturing of automotive sector components, it is typically observed that the dimensions of the machined product varies depending on the time at which the machining is performed. A similar or even more critical situation exists for die-mould machining as the machining time for the complete die or mould could span from a few days to a few weeks. It is therefore quintessential to understand the reason for the non-repeatability issue and literature suggests that one of the significant cause (upto 70%) can be traced back to thermal errors[J. B. Bryan, (1990) CIRP Annals - Manufacturing Technology, 39(2), 645–656; Ramesh et. al. (2000) The International Journal of Machine Tools and Manufacture, 40(9), 1257–1284]. Thermal errors are because of the change in the Tool Center Point (TCP) position which in turn could bebecause of various issues including the ambient temperature changes associated with day-night and seasonal cycle, the thermo-mechanical history of the machine and the in-process heat generation due to machining. It is for this reason that a few machine tool users workaround the thermal issues through the following strategies: first, through the commissioning of the machine in a temperature controlled room; second, through the manual perturbation of the reference position of TCP at regular intervals of time and finally through the use of a warm-up cycle of the machine in case the machine is switched ON after a prolonged shut-off period. All of the mentioned workaround strategies mentioned either increase the capital cost requirements or results in lower productivity. Compensation of thermal errors by developing models that are predicting the thermal behaviour of the machine tool is one of the most economical methods to take care of these issues. Chen et al. (2003), International Journal of Machine Tools & Manufacture (43), 1163–1170 calibrated the models considering the dynamic changes in the thermal behavior of the machine tool and suggested to develop displacement-based models instead of conventionally used temperature-based models. Although the adopted strategy of considering displacement-based model fares slightly better than the temperature-based models owing to the non-linearity of the thermal behavior, the complexity of it in real-time applications makes it difficult to deploy at the shop floor.
The conventional way of model building with temperature as the field variable is adopted by Liu et al. (2016) The International Journal of Machine Tools and Manufacture, (113), 35-48 with linear framework. Ridge regression fares better than the conventional least square linear models, but the calibration process of model building considering the eventual thermal error and all the temperatures together introduces robustness issues in the model.
To improve the predictive capability of the model, Yang et. al. (2014), Procedia CIRP (17) 698 – 703 and Yang et. al.(2015), The International Journal of Advanced Manufacturing Technology, (77), 1005–1017 tried to include the non-linear frameworks of neural network and Support Vector Machines (SVM) to calibrate the models. Although they improved the predictive capabilities of the models, the complexity introduced with higher computational time makes there practical implementation much more challenging.
US2013/0302180 has come up with a procedure of warm-up cycle for a CNC machine. Although, it leads to improved thermal stability and lesser thermal errors, it leads to addition of unproductive processingstep thereby making it commercially non-viable.
Thus, there is a need for a robust method to increase productivity as well as repeatability of a CNC machine throughautomatic thermal compensation. The present invention aims to reduce thermal errors by employing a decomposition-based compensation technique. This is executed by first byunderstanding of the thermal behaviour of the spindle and subsequent development and deployment of the compensation model considering the: thermo-mechanical history, ambient temperature variations and the in-process machining heat generated within the spindle unit.

SUMMARY OF INVENTION
The present invention provides a method of real time thermal error compensation for CNC machine; said method comprising steps of: a) calibrating the CNC machine to assess the thermal errors as a function of thermo-mechanical history, internal residual heat generation and ambient temperature; said calibration comprising, i) fixing a metrological fixture comprising plurality of displacement sensors on table of 3-axis vertical machining center of the machine and plurality of thermal sensors onto the CNC machine; ii) calibrating the thermal error as a function of thermo-mechanical history effects by maintaining the residual/internal heat effects and ambient temperature constant; iii) calibrating the thermal error as a function of residual heat effects by maintaining the thermo-mechanical history effects and ambient temperature constant; and iv)calibrating the thermal error as a function of ambient temperature by idling the machine; b) communicating calibration data to controller of the CNC machine through Internet of Things software to set baseline for thermal distortion and spindle offset;c) measuring the ambient temperature and rear bearing temperature of the spindle via PLC module during real time machining process periodically and communicating the measurement to controller of the CNC machine; andd) updating, summing the errors during real time machining and communicating to the CNC controller using Internet of Things software to compensate for the thermal error. The invention also provides a system for compensating real time thermal error in a CNC machine during machining, said system comprising- metrological fixture with sensors to assess displacement; thermal sensors to asses distortion due to thermo-mechanical history, internal residual heat generation and ambient temperature; Internet of Things platform communicating the thermal changes to controller of CNC machine and compensating error due to thermal changes during machining. Additionally, the invention provides a method of developing a workpiece by a CNC machine compensating the thermal error, said method comprising-a) measuring the ambient temperature and rear bearing temperature of the spindle via PLC module during real time machining process periodically and communicating the measurement to controller of the CNC machine; and b) updating, summing the errors during real time machining and communicating to the CNC controller using Internet of Things software to compensate for the thermal error.

BRIEF DESCRIPTION OF FIGURES.
The features of the present invention can be understood in detail with the aid of the appendedfigures. It is to be noted however, that the appended figures illustrate only typical embodimentsof this invention and are therefore not to be considered limiting of its scope for the invention.
Figure 1: Ambient temperature cycle for a day to find the approximate constant temperature slots.
Figure 2: The flow diagram of the thermal compensation algorithm deployment.
Figure 3: A schematic of the experimental set-up along with the precision metrology fixture containing the precision capacitive displacement sensors.
Figure 4:Comparison of thermal errors observed for a 29 hour experiment with and without compensation being applied.

DETAILED DESCRIPTION OF INVENTION
The present invention provides a method for developing a robust thermal compensation considering the major influencers of the thermal behaviour of a CNC machine, i.e., thermo-mechanical history effects, residual heat generation in spindle and ambient temperature variations, separately. The overall thermal error is split into various components (decompose) to build models individually and finally the results are superimposed to get overall thermal errors.
The foregoing description of the embodiments of the invention has been presented for the purpose of illustration. It is not intended to be exhaustive or to limit the invention to the precise form disclosed as many modifications and variations are possible in light of this disclosure for a person skilled in the art in view of the figures, description and claims. It may further be noted that as used herein and in the appended claims, the singular forms "a", "an", and "the" include plural reference unless the context clearly dictates otherwise. Unless defined otherwise, all technical and scientific terms used herein have the same meanings as commonly understood by person skilled in the art.
In one embodiment of present invention,CNC machine is selected from a group comprising vertical machining centre, horizontal machining centre, double column machining centre, turning machines, grinding machines and special purpose machine.
In another embodiment of present invention, the calibration of the thermal error as a function of thermo-mechanical history effects, comprising steps of:
measuring ambient temperature at predefined time slots at heated and cold conditions as well as the Tool Center Point (TCP) distortion using metrological fixture periodically and communicating the data to the controller; and
preparing the thermo-mechanical baseline module by transformation of displacement recorded by the metrologicalfixture to obtain thermal distortion, ?x, ?y,?z, ?R, pitch and yaw angles by the controller.
In yet another embodiment of present invention, the error as a function of thermo-mechanical effect is considered when the ratio of temperature between rear bearing of the spindle and ambient temperature is above threshold maximum or below threshold minimum depending on spindle type.
In yet another embodiment of present invention, the calibration of the thermal error as a function of residual heat effects, comprises steps of:
warming up the machine and running it alternatively at constant speed and random speed;
measuring ambient temperature and TCP distortion at predefined time slots using metrological fixture periodically and communicating the data to the controller; and
preparing the residual heat baseline module by transformation of the displacement recorded by metrologicalfixture to obtain thermal distortion, ?x, ?y,?z, ?R, pitch and yaw angles by the controller.
In yet another embodiment of present invention, the calibration of the thermal error as a function of ambient temperature, comprises steps of:
keeping the machine at idle state;
measuring ambient temperature and TCP distortion using metrological fixture periodically and communicating data to the controller; and
preparing the ambient temperature baseline module by transformation of the displacement recorded by metrology fixture to obtain thermal distortion, ?x, ?y, ?z, ?R, pitch and yaw angles by the controller.
preparing the ambient temperature baseline module by transformation of the displacement recorded by metrology fixture to obtain thermal distortion, ?x, ?y, ?z, ?R, pitch and yaw angles by the controller.
preparing the ambient temperature baseline module by transformation of the displacement recorded by metrology fixture to obtain thermal distortion, ?x, ?y,?z, ?R, pitch and yaw angles by the controller.
The present invention is in relation to a system for compensating real time thermal error in a CNC machine during machining, said system comprising- metrological fixture with sensors to assess displacement; thermal sensors to asses distortion due to thermo-mechanical history, internal residual heat generation and ambient temperature; Internet of Things platform communicating the thermal changes to controller of CNC machine and compensating error due to thermal changes during machining.
The present invention is also in relation to a method of developing a workpiece by a CNC machine compensating the thermal error, said method comprising-
measuring the ambient temperature and rear bearing temperature of the spindle via PLC module during real time machining process periodically and communicating the measurement to controller of the CNC machine; and
updating, summing the errors during real time machining and communicating to the CNC controller using Internet of Things software to compensate for the thermal error.

The global process flow is depicted in Scheme 1 which includes measuring critical parameters such as ambient and bearing temperatures, monitoring them, setting controlling parameters and communicating the required information (external tool and work offsets) to the CNC system (physical system). An Internet-of-Things (IoT) based system, IRIS, is used as the platform to set-up communication between different physical systems. It reads the real-time information from temperature sensors through the Programmable Logic Control (PLC) module and passes on the data to be used.

SCHEME-1
The detailed process for thermal compensation usingdecomposition-based approach is described for a C-frame vertical machining centre as an example.
The overall thermal error is divided into the individual effects of thermo-mechanical history, ambient variations and the internal/residual heat of the spindle.The objective of decomposing overall thermal error into function of individual component is to calibrate the mathematical functions of each effect separately. The overall thermal error is the result of superposition of mathematical functions of each effect. It is taken care during the calibration for a particular effect that the experiments are conducted such that the respective effect is predominant while others can be neglected. For instance, the thermo-mechanical history compensation model is calibrated by performing experiments wherein the spindle is started at different thermo-mechanical states while the other effects of ambient temperature variations and process heat variations are kept minimal. In order to minimize the effects of ambient variations, particular time-interval slots (Slot-1 and Slot-2) as shown in Figure 1 are selected to conduct the experiments while spindle is kept idle to minimize the internal heat effects. Similarly, during the calibration of residual compensation, the experiments are so conducted wherein the thermo-mechanical or the ambient temperature variations are kept at minimum. In other words, the compensation is developed in a decomposition-based manner and it is because of this methodology, the TCP distortion is predicted in a robust manner.
The details of the thermo-mechanical history modulebuilding process are as follows. The thermo-mechanical state of spindle is a reflection of the amount of stabilization or plateau reached in terms of evolution of temperature and thermal distortion. The extreme ends of the spectrum over which thermo mechanical state of a spindle can lie is from very cold to very hot. Thermo-mechanical state reached by the spindle is with respect to the ambient conditions. Hence, a simple workaround is by associating the internal state of the spindle in terms of temperature observed with respect to the ambient temperature. For instance, if the spindle is switched 'OFF' for a longer time interval, due to the material properties (the thermal conductivity of spindle), the bearing temperatures are relatively lower than the ambient temperature; while during spindle operation, the bearing temperatures are higher with respect to ambient due to friction because of rotation of spindle. Quantitatively, a metric is defined through the ratio of the bearing temperature to the ambient temperature which provides an idea about the initial thermal state. Although either front or rear bearing temperatures can be used as the internal variable of the spindle, typically front bearings are externally cooled through a recirculation cooler unit in a motorized spindle and therefore an inverse correlation exists between front bearing temperature and the axial thermal distortion at the TCP. Hence, temperature at rear bearing is considered as the reflection of the internal state of the spindle. An empirical ratio defined through k is used to identify the initial thermal state of the spindle and thereby to build predictive model:
k= (T_rb (t=0))/(T_a (t=0)) (Eqn 1)
where, T_rb (t=0) and T_a (t=0) are the rear bearing temperature and ambient temperature when the CNC machine tool is switched ON.
For calibration of model parameters, several experiments are conducted at different thermo-mechanical states spanning from extreme cold to extreme hot and the thermal growth of the spindle, ?z, is estimated using f(k). The quadratic function for f(k) provides the best fit and the compensation applied to cater to the thermo-mechanical history in an exponential fashion following the time constant of the spindle,t_s.
?(?z)?_th=?(1-exp???(-t/t_s ))? f(k) (Eqn 2)
The next sub-module is developed to cater to the residual thermal distortion at TCP. A single temperature sensor is employed on the outer race of rear bearing for compensation. The experiments wherein spindle speeds are varying with respect to time are conducted over such an interval wherein the ambient temperature does not change much and the thermo-mechanical histories are similar and the resulting data are used to train this model. This ensures, the training dataset will constitute only the effects due to internal heat sources and nullify the contribution of external heat source, i.e., ambient temperature variations. The experiments are intended to mimic real-life scenario and therefore involves wide-ranging spindle speeds. The model is then formulated and trained in a linear regression framework.
?(?z)?_r=a ?T_rb (Eqn 3)
The final step in the development of thermal compensation module is the training of ambient temperature module. It is carried out by examining the change in the TCP distortion of the spindle over day-night cycle with the spindle being idle and kept in a stand-alone position. It is important to note that the spindle fixed onto the machine tool and a stand-alone spindle behaves very differently because of ambient temperature changes. The main reason can be attributed to the fact that the CNC machine tool is substantially large compared to the spindle and is made up of parts which have very different thermal inertias (i.e., time constants) and they behave differently when subjected to a typical day-night cycle. This module is then formulated through the equation,
?(?z)?_a=ß ?T_a (Eqn 4)
??z?_tot= ?(?z)?_th+?(?z)?_r+?(?z)?_a (Eqn 5)
The process flow of deployment of the compensation algorithm is shown in Figure 2. The IoT device is switched ON automatically when the CNC machine tool is switched ON. It is important that the relative changes in temperature for the algorithm are from the proper reference time and hence the algorithm is automatically started after the work and tool offsets of the machining process and the origin are defined. It is to be noted that the operator can set multiple tool and work offsets as and when the need arises and the compensation algorithm takes them into account as well. The real-time inputs for the compensation algorithm are the ambient and rear bearing temperatures, which are acquired every ?t seconds using the PLC module and the IoT device named IRIS. The three sub-modules, ambient module (I), thermo-mechanical history module (II) and the residual heat module (III) are initialized at the same time (t=0). While the ambient and residual heat models are always in the scheme of things with the respective temperatures being recorded every ?t seconds, thermo-mechanical history module is only applied in case machine is not in the warmed-up state during the start of the machining process. A minimum and maximum threshold values(?) depending on spindle type are selected from the model to make sure that the effect of thermo-mechanical history is applied only when it is required. The summation of thermal expansion from the individual models is the total thermal error which as the final step is communicated to the CNC controller using the IoT device, IRIS to change the required offset.

Experimental:
Arrangements and process for physical measurements for model calibration:
The experimental set-up includes a metrology fixture fixed on the table of a 3-axis vertical machining center. The components of the machine tool shown in figure 3 are labeled in an ascending manner with numeral 1 to 7 are as follows: table [1], cross-slide [2], bed[3], column [4], milling head[5], spindle[6] and metrology fixture[7]. Resistance Temperature Detector (RTD) sensors[8] are used for ambient and rear bearing temperatures monitoring using a PLC. In addition, a precision ground disk [9] is rigidly clamped onto spindle shaft to act as its extension and a steel fixture is designed to place six precise capacitive sensorsso as to measure spindle distortion at six designated points (Figure 3; sensors Sa, Sb,Sc, Sd, Se and Sf) on the rotating disk. The three sensors each are used for radial and axial displacement measurements. The sensors are placed at 120 degrees angle from each other while the radial sensor S_a is placed at angle ? from X-axis (Figure 3). The displacements recorded from sensors (Sa, Sb,Sc, Sd, Se and Sf) are then transformed to obtain thermal distortion: ?x, ?y,?z, ?R, pitch and yaw angles respectively (?x and ?y) using the solution of following matrix equations:
[¦(1&R_0 sin?&-R_0 cos?@1&?-R?_0 sin?(60+?)&R_0 cos?(60+?)@1&R_0 cos?(30+?)&R_0 sin?(30+?))][¦(?z@?_x@?_y )] =[¦(S_d@S_e@S_f )], (Eqn. 6)
[¦(cos?&sin?&1@-cos?(60+?)&-sin?(60+?)&1@-sin?(30+?)&cos?(30+?)&1)][¦(?x@?y@?R)] = [¦(S_a@S_b@S_c )] (Eqn. 7)
Where, R_0 is the radius of the disk on which radial displacement measurements are performed (Figure 3). Custom-built IoT system named IRIS is used to synchronize the data from CNC controller (such as spindle load and motor temperature) along with temperature data and simultaneously, the spindle distortion is calculated through capacitive sensors (Equations 6 and 7). For the set of spindles used in the examples, threshold values of 0.85 and 1.15 for minimum and maximum values arefound to be suitable.
Calibration of process parameters
The experiments conducted for model calibration of each effect are planned in such a way that the contributions of other two effects are minimal on the overall thermal errors as discussed before. For instance, the thermo-mechanical experiments are conducted at the time slots of day, generally of about 2 hours, when the variations in ambient temperature are minimal. In order to simulate the extreme conditions of thermo-mechanical effect and collect a large data set, the initial state of the spindle is varied from extreme cold to extreme hot conditions. While the extreme hot condition is simulated by running the spindle at around 15000 rpm for more than an hour before displacement and temperature data collection, the extreme cold situation is simulated by running cooled coolant oil in the spindle.
The calibration of model for residual heat effect is also done at aforementioned time slots of the day to keep the ambient effect minimal. Also, it is made sure that the experiments are done with a warmed up state of the machine to ensure minimal thermo-mechanical effects on thermal errors. The experiments done are consisting of constant speed throughout the timeline as well as random speeds to take care of the large fluctuations observed in real-time machining.
As far as ambient module is considered, the displacement and temperature data are collected over a few days (2 to 4 days) with spindle in idle condition. The thermo-mechanical and residual heat effects do not come into existence for this experiment and the thermal errors are only a function of ambient temperature variations. The combinations of these models are used to predict the overall thermal errors at the TCP.
Validation of the compensation process
In order to validate the mentioned methodology, a validation experiment is conducted for more than 24 hours as it reflects typical time spent on a manufacturing process. The whole experiment is planned in a way that during different regions of the timeline, one effect is predominant. Also during some other time intervals the thermal errors are a combination of all the effects. Figure 4 shows the validation experiment along with the displacement observed with and without compensation algorithm applied. It can be easily observed that the error which is more than 45µm when the compensation is not applied reduces to less than 18µm after applying compensation.
The process of this invention utilizes only two temperature sensors (rear bearing and ambient) to build and deploy the models, the amount of thermo-mechanical history effect present is measured empirically and quantitative conversion to thermal error is done using time constant of the spindle. The communication setup of IRIS with CNC is independent of the controller and hence, thermal compensation technique works for all type of controllers and consequently for different kinds of CNC machine tools: vertical machining centre, horizontal machining centre, double column machining centre, turning machines, grinding machines and special purpose machine tools. The method is downward compatible, i.e., the compensation methodology can be deployed on existing CNC machine tools as well as new machine tools.