TITLE OF INVENTION Power converter | FIELD OF THE INVENTION j 5 {0001} I The present invention relates to a power converter, and more particularly to a three-phase power converter j that is formed with three clusters, each of which is a F serial body including one or a plurality of unit cells. ! 10 I BACKGROUND OF THE INVENTION f {0002} A cascade multi-level converter (CMC) is a circuit f I system that can output voltages equal to or higher than | 15 the withstand voltages of switching elements used » therein; the switching elements are, for example, I insulated gate bipolar transistors (IGBTs) or other i switching elements that can be turned on and off in a f controlled manner. According to NPL 1, the CMC is 20 formed by star-connecting three serial bodies, each of ( which is formed with a reactor and a cluster, the ! cluster being a serial body formed with a plurality of I I unit cells. f {0003} ( 25 According to NON-PATENT LITERATURE (NPL) 1, each I I. - 3 - I unit cell is a single-phase full bridge circuit; it has t a plurality of switching elements and a DC capacitor. | The unit cell turns on and off the relevant switching I element in a controlled manner to output a voltage 5 across the DC capacitor (referred to below as the DC voltage), a voltage with a polarity opposite to the polarity of the DC voltage, or a zero-voltage. i {0004} j Since each cluster is a serial body having one or a j ) 10 plurality of unit cells, the output voltage of the j cluster (referred to below as the cluster voltage) is j I the output voltage of the one unit cell included in the ! cluster or the sum of the output voltages of the ! plurality of unit cells therein. When each cluster j 15 includes a plurality of unit cells, if the switching j i timing of each unit cell in the cluster is i appropriately shifted, the cluster voltage can have a multi-level waveform. Accordingly, the harmonic j component of the cluster voltage can be reduced by I 20 increasing the number of unit cells included in each | j cluster. I {0005} NPL 1 indicates experimental results of a CMC-based static var compensator (referred to below as the CMC- 25 STATCOM) that has a CMC linked to a power supply system. I f - 4 - I NPL 1 also discloses that even if an instantaneous i! voltage drop is caused in the power supply system, the l operation of the CMC-STATCOM can be continued. l 5 PRIOR TECHNICAL LITERATURE [ NON-PATENT LITERATURE (NPL) ; {0006} I NPL 1: Yoshii, Inoue, Akagi "Transuresu kasukedo ] PWM STATCOM no chokuryudenatsuseigyoho no kento (Zero- [ 10 sequence voltage control of a cascade PWM converter ! with star-configuration)" lEE-SPC and lEE-IEA materials, ] SPC-07-115/IEA-07-38, pp. 32-36 ; SUMMARY OF THE INVENTION 15 {Technical Problem} {0007} NPL 1 discloses DC voltage balancing control by which the DC voltages of a plurality of unit cells included in the CMC are balanced (simply referred to 20 below as balancing control). ; {0008} Balancing control is classified into cluster balancing control and inter-stage balancing control. {0009} 25 Cluster balancing control is control by which the I I I f I • t i t i i ! "^ i I - 5 - i j I ! average of the DC voltages of the unit cells included | i I I in each cluster (referred to below as the cluster l i i j average DC voltage) is made to match the average of the | I I i DC voltages of all unit cells (referred to below as the { j 5 total average DC voltage) . | i I i {0010} j i I i In cluster balancing control, a zero-phase voltage | ! command value calculated from a difference between the | ; total average DC voltage and the cluster average DC | ' t j 10 voltage is superimposed on a cluster voltage command i \ j value to be supplied to each cluster to adjust active ! i electric power that flows into the cluster. Thus, j j feedback control is performed so that the difference j between the total average DC voltage and the cluster j 15 average DC voltage becomes zero. j i {0011} ! In inter-stage balancing control, the DC voltages 1 of a plurality of unit cells in one cluster are | balanced. i 20 {0012} j According to NPL 1, even if an instantaneous 1 voltage drop is caused in the power supply system, the I operation of the CMC-STATCOM can be continued, as I j described above. Even when the balancing control j I 25 described above is carried out, however, the cluster j i i j j j i I - 6 - [ average DC voltage becomes unbalanced at the occurrence I of an instantaneous voltage drop. That is, the DC voltage is increased or decreased from the rated value. f {0013} [ 5 Accordingly, the CMC disclosed in NPL 1 needs to be designed so that DC capacitors withstand a high voltage i against a DC voltage rise at the occurrence of an I instantaneous voltage drop, resulting in large DC i voltage capacitors. j 10 {Solution to Problem} j {0014} j First, the reason why the CMC-STATCOM described in j NPL 1 causes a cluster average DC voltage imbalance at j the occurrence of an instantaneous voltage drop will be j i 15 described below. A solution to prevent the cluster ! average DC voltage imbalance will then be described. i {0015} j Voltage in the power supply system (simply referred j to below as the power supply system voltage) at the j 20 occurrence of an instantaneous voltage drop includes a | negative-phase sequence component. While an instantaneous voltage drop continues, each cluster j outputs a negative-phase sequence voltage that is nearly equal to the negative-phase sequence component 25 included in the power supply system voltage to prevent i j I f ! i - 7 - I negative-phase sequence current from flowing. i {0016} I Since negative-phase sequence voltage output from I each cluster and positive-phase sequence reactive l 5 current flowing in the cluster generate unbalanced t { electric power, active electric power that flows into I the cluster becomes unbalanced. As a result, the cluster average DC voltage becomes unbalanced. I {0017} 10 In cluster balancing control in NPL 1, after an j imbalance of the cluster average DC voltage has been j detected, a zero-phase voltage command value that j balances the cluster average DC voltage is calculated. l Since a low-pass filter (LPF) is used during the { i 15 detection of the DC voltage of each unit cell, however, the response speed in balancing control is substantially determined by the LPF. Therefore, i phenomena in several tens of milliseconds to hundreds of milliseconds cannot be followed. This makes it j 20 impossible to prevent a rise or drop of the DC voltage. | {0018} i In the CMC, in which a unit cell is a single-phase converter, the DC voltage varies at twice the frequency of the power supply system. To eliminate this variation 25 in the detection and control of the DC voltage, the LPF I f - 8 - J is indispensable. I {0019} To address the above problem, the present invention j provides a solution described below. 5 {0020} i In a power converter that has a star-connected cascade multi-level converter (CMC) linked to a power supply system, the CMC being formed by star-connecting j serial bodies, each of which is formed with a reactor j 10 and a cluster, the cluster being a serial body formed j i with one or a plurality of unit cells, the present invention provides a power converter characterized in i that a potential at a point at which three serial bodies, each having the reactor and the cluster, are 15 star-connected is controlled so that the potential varies at the same amplitude as the amplitude of the negative-phase sequence voltage of the power supply system. {0021} 20 In a power converter that has a star-connected cascade multi-level converter (CMC) linlced to a power supply system, the CMC being formed by star-connecting serial bodies, each of which is formed with a reactor and a cluster, the cluster being a serial body formed 25 with one or a plurality of unit cells, the present i I I: - 9 - j invention provides a power converter characterized in j that when the power supply system causes an j instantaneous voltage drop, an amplitude by which the ; potential at the point at which the three serial bodies ! 5 are star-connected varies is made substantially I constant from a half cycle after the occurrence of the j instantaneous voltage drop until the instantaneous voltage drop ends. ; {0022} i 10 In a power converter that has a star-connected j cascade multi-level converter (CMC) linked to a power i supply system, the CMC being formed by star-connecting serial bodies, each of which is formed with a reactor j i and a cluster, the cluster being a serial body formed i 15 with one or a plurality of unit cells, the present invention also provides a power converter characterized in that if the root-mean-square value and phase angle of the negative-phase sequence component included in a voltage of the power supply system are respectively 20 denoted V2 and (|)2 and the phase angle of the positivephase sequence component of a current that flows into the reactor is denoted 51, a zero-phase voltage having root-mean-square value VO and phase angle (|)0, which are substantially represented by the equations below, is 25 added to an output voltage command value to be supplied I ^ i - 10 - ( f to each cluster. ? f {Eq. 1} j VO = V2 I {Eq. 2} j 5 (|)0 = 2 X 51 - (|)2 ± 71 I The present invention also provides a power I converter characterized in that moving average { i calculation in which a half cycle of the frequency of t the power supply system or its integer multiple is | 10 handled as a time window is used in the detection of j I root-mean-square value V2 and phase angle (|)2 of the j j negative-phase sequence component included in the j voltage of the power supply system and in the detection i of phase angle 51 of the positive-phase sequence j 15 current of the reactor. j j {0023} I In a power converter that has a star-connected j cascade multi-level converter (CMC) linked to a power | supply system, the CMC being formed by star-connecting i 20 serial bodies, each of which is formed with a reactor | and a cluster, the cluster being a serial body formed i with one or a plurality of unit cells, the present I invention also provides a power converter characterized i in that when the power supply system causes an ! 25 instantaneous voltage drop while the power converter is | j i I 'i I I: -i - 1 1 - } outputting a negative-phase sequence current, a zero- f phase voltage output by the cluster is made substantially constant from a half cycle after the t occurrence of the instantaneous voltage drop until the 5 instantaneous voltage drop ends. t {0024} In a power converter that has a star-connected I cascade multi-level converter (CMC) linked to a power | supply system, the CMC being formed by star-connecting { I 10 serial bodies, each of which is formed with a reactor and a cluster, the cluster being a serial body formed j with one or a plurality of unit cells, the present i invention also provides a power converter characterized j ! in that if the root-mean-square value and phase angle 15 of the positive-phase sequence component included in 1 S the voltage of the power supply system are respectively j denoted VI and ^1, the root-mean-square value and phase ! j angle of the negative-phase sequence component included in the voltage of the power supply system are I 20 respectively denoted V2 and and positive-phase sequence q-axis current . The feedforward calculator 309 also obtains positive-phase sequence current root-mean-square value II and 25 positive-phase sequence current phase angle 51 from the I I t, r I i -i f I - 23 - [ sizes and amplitudes of vectors [, ] through an I orthogonal coordinate-polar coordinate converting block f 403. } {0059} I 5 The feed-forward calculator 309 also uses a ! negative-phase sequence dq converting block 401 to perform negative-phase sequence dq conversion on phase j i voltages VSu, VSv, and VSw of the power supply system j 101, and calculates a moving average through the moving j 10 average calculator 402, in which a half cycle of the j power supply system is handled as a time window, i obtaining negative-phase sequence d-axis voltage and negative-phase sequence q-axis current . The I feed-forward calculator 309 also obtains negative-phase 15 sequence voltage root-mean-square value V2 and negative-phase sequence voltage phase angle <|)2 from the sizes and amplitudes of vectors [, ] through the orthogonal coordinate-polar coordinate converting block 403. 20 {0060} The inventors found from II, 51, V2, and (j)2 that active electric power that flows into the clusters 104 to 106 can be made zero by superimposing the zero-phase feed-forward term VOFF*, which has root-mean-square 25 value VO and phase angle (|)0 represented by the I f I I" I I i - 24 - equations below, on cluster voltage command values Vu*, Vv*, and Vw*. The mechanism of the present invention ! I will be described later. { {Eq. 6} j 5 VO = -V2 j { {Eq. 7} j (|)0 = 2 X 51 - (t)2 I A sine wave generator 404 generates the zero-phase { feed-forward term VOFF* as a sine wave signal having 10 root-mean-square value VO and phase angle ^0. The f I obtained VOFF* is superimposed on cluster voltage | command values Vu*, Vv*, and Vw*. } {0061} I When each cluster is controlled as described above, | 15 the potential at the point M in FIG. 1 varies with an | amplitude that is almost the same as the amplitude of } the negative-phase sequence component included in the I power supply system 101. {0062} 20 Effects obtained by the present invention will be | described with reference to FIGs. 5 and 6. } I {0063} FIG. 5 shows schematic waveforms in individual members in a case in which the prior art has been used. 25 The waveforms at the top are schematic waveforms of I i I I •i I i - 25 - phase voltages VSu, VSv, and VSw of the power supply system 101, followed by schematic waveforms of currents lu, Iv, and Iw that flow into the reactors 103, schematic waveforms of DC voltages VCuj, VCvj, and VCwj, 5 schematic waveforms of DC voltages VCufj, VCvfj, and VCwfj observed after they have passed through the LPFs 301, and a schematic waveform of the zero-phase feedback term VOFB* (j = 1, 2, ..., N) in that order. {0064} 10 In FIG. 5, the horizontal axis indicates time or phase angle, and the vertical axis indicates amplitudes of voltage and current in an arbitrary unit (a.u.). {0065} The prior art lacks the feed-forward calculator 309 15 shown in FIG. 3. {0066} FIG. 5 shows a case in which a 50% instantaneous I voltage drop has been generated positive-phase sequence I I u. While the instantaneous voltage drop continues, VSu, I I 20 VSv, and VSw include both a positive-phase sequence | I component and a negative-phase sequence component. f {0067} I I To prevent negative-phase sequence currents from | i flowing, the clusters 104 to 106 output negative-phase I 25 sequence voltages having almost the same amplitude and t I I I { I f i I i I I I H I i I ! ^ I - 26 - J ! phase angle as the negative-phase sequence components j i of VSu, VSv, and VSw. That is, cluster voltages Vu, Vv, ! and Vw include negative-phase sequence components. I {0068} I 5 Since the negative-phase sequence components i included in Vu, Vv and Vw and lu, Iv and Iw including f I • I I only the positive-phase sequence components form i j unbalanced electric power, active electric power that { 1 flows into each cluster and active electric power i 10 output from the cluster are unbalanced. As shown in FIG. j 5, therefore, DC voltages VCuj, VCvj, and VCwj are I j unbalanced. i {0069} I The cluster balancing controller 308 calculates the | 15 zero-phase feedback term VOFB* so that VCufj, VCvfj, j and VCwfj, which are DC voltages that have passed i j through the LPFs 301, are balanced. As VCufj, VCvfj, j and VCwfj are more unbalanced, therefore, the amplitude ! of VOFB* is also increased. I I I 20 {0070} I I The amplitude of VOFB* converges to a fixed value I only after the imbalance among VCuj, VCvj, and VCwj has reached a certain value. At that time, active electric power that flows into each cluster becomes zero. 25 {0071} j i I < i I - 27 - In FIG. 3, DC voltages VCuj, VCvj, and VCwj are detected after they have passed through the LPFs 301, | so the response characteristic of the cluster balancing controller is substantially determined by the LPF 301. | 5 Furthermore, since each unit cell 107 is a single-phase full bridge and variations that are caused in DC j voltages with a frequency twice the frequency of the power supply system are removed in control and calculation, the LPF 301 is indispensable. In the prior j 10 art, therefore, it has been impossible to prevent VCuj, ! VCvj, and VCwj from being unbalanced. j {0072} j FIG. 6 shows schematic waveforms in individual { i members in a case in which control in the present j 15 invention has been carried out. The waveforms at the top are schematic waveforms of phase voltages VSu, VSv, and VSw of the power supply system 101, followed by I schematic waveforms of currents lu, Iv, and Iw that flow into the reactors 103, schematic waveforms of DC 20 voltages VCuj, VCvj, and VCwj, schematic waveforms of DC voltages VCufj, VCvfj, and VCwfj observed after they have passed through the LPFs 301, and a schematic waveform obtained as the sum of the zero-phase feedback term VOFB* and the zero-phase feed-forward term VOFF* 25 (VOFB* + VOFF*) in that order. r % ! t }- < I - 28 - I {0073} j In FIG. 6, the horizontal axis indicates time or f phase angle, and the vertical axis indicates amplitudes S of voltage and current in an arbitrary unit (a.u.). ] 5 {0074} i Unlike the prior art in FIG. 5, the amplitude of i the sum of VOFB* and VOFF* shown at the bottom in FIG. 1 6 has become substantially constant immediately after j the occurrence of the instantaneous voltage drop. It is i 10 found that due to this, DC voltages Vcu, Vcv, and Vcw | shown as the third graph from the top in FIG. 6 also j caused almost no change before and after the i instantaneous voltage drop. I {0075} i 15 The mechanism of the present invention will be I described below. {0076} When the zero-phase feed-forward term VOFF* obtained by the feed-forward calculator 309 in FIG. 3 20 is superimposed on cluster voltage command values Vu*, Vv*, and Vw*, a phase difference between Vu and lu, a phase difference between Vv and Iv, and a phase difference between Vw and Iw all become approximately 90 degrees, starting immediately after the occurrence 25 of an instantaneous voltage drop. That is, active f I I ! I - 29 - I electric power that flows into each cluster and active i f electric power output from the cluster become 0. I {0077} j I Since active electric power that flows into each j I 5 cluster and active electric power output from the f } cluster become 0, it becomes possible to suppress the f DC voltages from being unbalanced. J {0078} } In the above description, a case has been described i 10 in which the negative-phase sequence component of the i j system voltage is known immediately after the | occurrence of the instantaneous voltage drop. If, for example, a time equivalent to a half cycle of the power supply system is required in the detection of the I 15 negative-phase sequence component of the power supply 1 system, however, VOFF* becomes substantially fixed from I I a half cycle after the occurrence of the instantaneous I voltage drop until the instantaneous voltage drop ends. {0079} I 20 In the present invention, the zero-phase feedback f term VOFB* plays a role of suppressing the DC voltages f from being unbalanced due to variations in the I characteristics of parts in use and other factors. {0080} 25 With the power converter 102 in this embodiment, I < I I - 30 - j the clusters 104 to 106 are connected to the power j supply system 101 through their respective reactors 103. j I {0081} [ The present invention can also be applied to a case I 5 in which the clusters 104 to 106 are each connected to j the power supply system 101 through a transform instead of the reactor 103. {0082} The present invention can also be applied to a case j t I 10 in which the clusters 104 to 106 are each connected to | £ the power supply system 101 through both the rector 103 | and a transform. | {Embodiment 2} | I {0083} { 15 A second embodiment of the present invention will | f be described below. f f {0084} I In the second embodiment, the DC capacitor in each unit cell is made compact when compared with the CMCI 20 STATCOM in the prior art, as in the first embodiment. I I {0085} f I In the second embodiment, even if an instantaneous 1 I voltage drop occurs in the power supply system 101 in a I f state in which currents lu, Iv, and Iw flowing in the [ 25 reactors 103 include negative-phase sequence components, [ I I I I I I i |; I I I - 31 - f it is possible to prevent the DC voltage from being I unbalanced. { i {0086} J In the second embodiment, the feed-forward | I 5 calculator 309 in the first embodiment is replaced with | j a negative-phase sequence current ready feed-forward J i calculator 701 shown in FIG. 7. The negative-phase I I f I sequence current ready feed-forward calculator 701, | which is a difference between the first embodiment and j 10 the second embodiment will be described below with reference to FIG. 7. {0087} The negative-phase sequence current ready feedforward calculator 701 obtains positive-phase sequence 15 d-axis current and positive-phase sequence q-axis I current from currents lu, Iv, and Iw through the I I I positive-phase sequence dq converting block 305 and j j moving average calculator 402. The negative-phase | I sequence current ready feed-forward calculator 701 also I i 20 obtains positive-phase sequence current root-mean- i square value II and positive-phase sequence current phase angle 61 from the sizes and amplitudes of vectors | I [, ] through the orthogonal coordinate-polar I I coordinate converting block 403. I I 25 {0088} I t ! I ! I - 32 - I The negative-phase sequence current ready feedforward calculator 701 obtains negative-phase sequence } d-axis current and negative-phase sequence q-axis current from currents lu, Iv, and Iw through the 5 negative-phase sequence dq converting block 401 and moving average calculator 402. The negative-phase sequence current ready feed-forward calculator 701 also obtains negative-phase sequence current root-meansquare value 12 and negative-phase sequence current j I 10 phase angle 52 from the sizes and amplitudes of vectors [, ] through the orthogonal coordinate-polar j coordinate converting block 403. 1 {0089} j ! The negative-phase sequence current ready feed- | 15 forward calculator 701 obtains positive-phase sequence i d-axis voltage and positive-phase sequence q-axis voltage from voltages VSu, VSv, and VSw through the positive-phase sequence dq converting block 305 and i moving average calculator 402. The negative-phase 20 sequence current ready feed-forward calculator 701 also obtains positive-phase sequence voltage root-meansquare value VI and positive-phase sequence voltage phase angle ^1 from the sizes and amplitudes of vectors [, ] through the orthogonal coordinate-polar 25 coordinate converting block 403. i ^ J i - 33 - I i {0090} I I I The negative-phase sequence current ready feed- forward calculator 701 obtains negative-phase sequence I I d-axis voltage and negative-phase sequence q-axis } \ 5 voltage from voltages VSu, VSv, and VSw through i the negative-phase sequence dq converting block 401 and i 1 moving average calculator 402. The negative-phase i sequence current ready feed-forward calculator 701 also obtains positive-phase sequence voltage root-mean- I 10 square value V2 and positive-phase sequence voltage j phase angle ^2 from the sizes and amplitudes of vectors I [ [, ] through the orthogonal coordinate-polar ] coordinate converting block 403. {0091} 15 The inventors found that even when lu, Iv, and Iw include negative-phase sequence components, active electric power that flows into each cluster and active electric power output from the cluster can be made zero by superimposing the zero-phase feed-forward term VOFF*, 20 which includes root-mean-square value VO and phase angle ^Q represented by the equations below, on cluster voltage command values Vu*, Vv*, and Vw*. {Eq. 8} VO = -(VI X 12 X cos((t)l - 52) + V2 X II X cos ( and positive-phase sequence q-axis current . The negative-phase sequence current output-type feed-forward calculator 20 802 also obtains positive-phase sequence current rootmean- square value II and positive-phase sequence current phase angle 51 from the sizes and amplitudes ofvectors [, ] through the orthogonal coordinatepolar coordinate converting bloclc 403. 25 {0108} - 39 - [ The negative-phase sequence current output-type I feed-forward calculator 802 also uses the positive- f phase sequence dq converting block 305 to perform i positive-phase sequence dq conversion on phase voltages I 5 VSu, VSv, and VSw of the power supply system 101, and f I calculates a moving average through the moving average i I calculator 402, in which a half cycle of the power ! supply system is handled as a time window, obtaining j negative-phase sequence d-axis voltage and I I I 10 negative-phase sequence q-axis current . The S negative-phase sequence current output-type feedforward calculator 802 also obtains negative-phase sequence voltage root-mean-square value VI and i negative-phase sequence voltage phase angle ^1 from the 15 sizes and amplitudes of vectors [, ] through the orthogonal coordinate-polar coordinate converting block 403. {0109} The negative-phase sequence current output-type 20 feed-forward calculator 309 also uses the negativephase sequence dq converting block 401 to perform negative-phase sequence dq conversion on phase voltages VSu, VSv, and VSw of the power supply system 101, and calculates a moving average through the moving average 25 calculator 402, in which a half cycle of the power -A - 40 - supply system is handled as a time window, obtaining negative-phase sequence d-axis voltage and negative-phase sequence q-axis current . The negative-phase sequence current output-type feed- 5 forward calculator 802 also obtains negative-phase sequence voltage root-mean-square value V2 and negative-phase sequence voltage phase angle ^2 from the sizes and amplitudes of vectors [, ] through the orthogonal coordinate-polar coordinate converting 10 block 403. I {0110} 5 A phasor calculator 901 calculates negative-phase I sequence current root-mean-square value 12 and j negative-phase sequence current phase angle 52, which i f 15 are represented by the equations below, by using II, 61, ? VI, (|)1, V2, and ^2 . } {Eq. 10} I 12 = (V2/V1) X II {Eq. 11} i 20 62 = <|)1 + (1)2 - 51 + 71 j The phasor calculator 901 calculates the negative- I f phase sequence current command value feed-forward terms ! Id2FF* and Iq2FF* from obtained 12 and 52 through a polar coordinate-orthogonal coordinate converting block S 25 902. S i A - 41 - {0111} The internal structure of the negative-phase sequence current output-type cluster balancing controller 803 will be described with reference to FIG. 5 10. {0112} ; The negative-phase sequence current output-type cluster balancing controller 803 calculates AVCu, AVCv, and AVCw, which are respectively differences between < 10 given total average DC voltage VC and cluster average s DC voltages VCu, VCv and VCw. The negative-phase I sequence current output-type cluster balancing f controller 803 then performs aP conversion on the obtained AVCu, AVCv, and AVCw by using an aP converting I 15 block 1001 and obtains [AVCa, AVCP] a.s vectors on an aP j axis. The negative-phase sequence current output-type I cluster balancing controller 803 further calculates the f sizes Ave and amplitude t, of vectors [AVCa, AVCp] through an orthogonal coordinate-polar coordinate j 20 converting block 1002. I {0113} } The negative-phase sequence current output-type j cluster balancing controller 803 also calculates t amplitude l + 51 - 52) + 12 x sin((t)l - 2 X 52)) - V2 x II x (12 x sin((|)2 - 52 + 51) + 5 II x sin((|)2 - 2 X 6l)))/(Vl x 12 x (II x cos ((|)1 + 51 - 52) - 12 X cos((t)l - 2 X 52)) - V2 x II x (12 x cos((t)2 - 62 + 51) - II X cos ((1)2 - 2 X 51)))) 7. The power converter according to Claim 1, 2, or 6, wherein moving average calculation in which a half 10 cycle of a frequency of the power supply system or an integer multiple of the half cycle is handled as a time window is used in detection of root-mean-square value VI and phase angle (|)1 of the positive-phase sequence component included in the voltage of the power supply 15 system, in detection of root-mean-square value V2 and phase angle ^2 of the negative-phase sequence component included in the voltage of the power supply system, in detection of root-mean-square value II and phase angle 51 of the positive-phase sequence component of the 20 current that flows into the reactor, and in detection of root-mean-square value 12 and phase angle 52 of the negative-phase sequence component of the current that flows into the reactor. 8 . A power converter that has a star-connected 25 cascade multi-level converter (CMC) linked to a power - 49 - supply system, the CMC being formed by star-connecting serial bodies, each of which is formed with a reactor and a cluster, the cluster being a serial body formed with one or a plurality of unit cells, wherein a 5 negative-phase sequence component of a current that flows into a serial body formed with the cluster and the reactor is controlled so that the negative-phase sequence component varies at the same amplitude as an amplitude of a negative-phase sequence voltage included 10 in the power supply system. 9. The power converter according to Claim 8, wherein when the power supply system causes an instantaneous voltage drop, an amplitude by which the negative-phase sequence component of the current that 15 flows into the serial body formed with the cluster and the reactor varies is made substantially constant from a half cycle after occurrence of the instantaneous voltage drop until the instantaneous voltage drop ends. 10. The power converter according to Claim 8 or 9, 20 wherein if a root-mean-square value and a phase angle of a positive-phase sequence component included in the voltage of the power supply system are respectively denoted VI and ^1, a root-mean-square value and a phase angle of a negative-phase sequence component included 25 in the voltage of the power supply system are - 50 - respectively denoted V2 and ^2, and a root-mean-square value and a phase angle of a positive-phase sequence component of a current that flows into the reactor are respectively denoted II and 51, a negative-phase 5 sequence current having root-mean-square value 10, substantially represented by {Eq. 5}, and a phase angle 52, substantially represented by {Eq. 6}, is controlled so that the negative-phase sequence current flows into the serial body formed with the cluster and the reactor. 10 {Eq. 5] 12 = {V2/V1) X II {Eq. 6} 62 = (|)1 + (t)2 - 51 ± 71 11. The power converter according to any one of 15 Claims 8 to 10, wherein a cluster-specific average of voltages across DC capacitors in one or more unit cells included in one cluster is calculated for each cluster and control is performed so that a negative-phase sequence current by which electric power flows into a 20 cluster in which the cluster-specific average is smallest flows into each cluster. 12 . The power converter according to any one of Claims 8 to 10, wherein moving average calculation in which a half cycle of a frequency of the power supply 25 system or an integer multiple of the half cycle is -A - 51 - handled as a time window is used in detection of rootmean- square value VI and phase angle (|)1 of the positive-phase sequence component included in the voltage of the power supply system, in detection of 5 root-mean-square value V2 and phase angle ^2 of the negative-phase sequence component, and in detection of phase angle 51 of the positive-phase sequence current of the reactor. 13. The power converter according to any one of 10 Claims 1 to 12, wherein the unit cell is a single-phase full bridge. 14. The power converter according to any one of Claims 1 to 12, wherein a transformer is connected instead of the reactor. 15 15. The power converter according to any one of Claims 1 to 12, wherein a transformer is connected in series with the reactor.