GB2623189A - A High-Precision Model Predictive Current Control System and Method for a Dual Three-Phase Permanent Magnet Motor - Google Patents

Abstract

A predictive current control system for a dual three-phase permanent magnet motor uses a high-precision model. The system hardware comprises a dual three-phase permanent magnet motor, a direct current (DC) power supply, a pulse width modulation (PWM) module, an inverter, a position sensor, and a current sensor. System software comprise a synthesized 24 virtual voltage vector module, a speed controller, a coordinate transformation module, a delay module, a prediction module, a duty cycle calculation module, a simplification module, and a cost function module. By expanding the conventional 12 virtual voltage vectors to 24 virtual voltage vectors with equal amplitude and uniform phase angles, control accuracy is improved without compromising voltage utilization efficiency. A duty cycle calculation method based on minimum error enables simultaneous tracking of the d-axis and q-axis currents even in the case of a single effective virtual voltage vector, ensuring optimal duty cycle output.

Description

A HIGH-PRECISION MODEL PREDICTIVE CURRENT CONTROL SYSTEM AND METHOD FOR A DUAL THREE-PHASE PERMANENT MAGNET MOTOR

Technical Field

The present invention belongs to the field of predictive control technologies for multiphase motors, and particularly relates to a high-precision model predictive current control system and a method for a dual three-phase permanent magnet motor.

Background

With the rapid development of high-end sectors such as transportation, aerospace, and defense industries, there is a growing demand for further improvement in motor systems, which serve as vital components of equipment. Multiphase permanent magnet motors are preferred for advanced motor systems due to their advantages such as high power density, high efficiency, and excellent fault tolerance capabilities. Among these motors, the dual three-phase permanent magnet motor, featuring a special structure with center point isolation and two sets of windings connected with a phase shift of 30 degrees, has gained wide application by eliminating the 6th-order torque ripple. Model predictive control strategy exhibits excellent performance in power converter applications, owing to its advantages of multivariable control, ease in handling nonlinear constraints, and intuitive implementation. However, it has drawbacks such as high computational requirements and high torque ripple. In a Chinese patent titled "Low-Computational Model Predictive Torque Control Method for Dual Motor Series System" (Patent No.: 202110774817.4), a low-computational model predictive control method is disclosed, which reduces the computational load by calculating the cost functions of only two voltage vectors. However, this method necessitates the calculation of the position of the reference voltage vector, leading to increased system complexity by incorporating two observers. Another Chinese patent titled "Model Predictive Control Method for Reducing PMSM Torque Ripple and Flux Ripple" (Patent No.: 202210499366.2) presents a method to reduce torque and flux ripple in permanent magnet synchronous motors by widening the modulation range using multiple voltage vectors within a single period. Although it achieves some effect, the computation involved is complex. When applying the model predictive control algorithm to the field of multiphase motors, the number of candidate voltage vectors increases exponentially, resulting in higher computational requirements. Moreover, multiphase motors include harmonic subspaces that must be controlled during system operation to avoid negative impacts on motor performance and significant losses. Therefore, in order to enhance the application of model predictive control in the field of multiphase motors, there is an urgent need to conduct relevant research to reduce the computational burden of the algorithm or explore techniques related to improving torque and flux ripple.

Summary

The objective of the invention: To address the issues of high torque ripple and heavy computational burden in model predictive control for dual three-phase permanent magnet motors, a redesign of the control set is proposed. By expanding the conventional 12 virtual voltage vector control set, a set of 24 virtual voltage vectors with equal magnitudes and evenly distributed phase angles are designed to improve control accuracy without sacrificing voltage utilization efficiency. Furthermore, a minimum error-based duty cycle calculation method is introduced, enabling simultaneous tracking of d-axis and q-axis currents and ensuring optimal output duty cycle even in the case of a single effective virtual voltage vector. Additionally, the process of traversing all voltage vectors in the prediction control is simplified, reducing the computational burden of the algorithm. This invention significantly enhances the accuracy of model predictive control by expanding the control set and reducing duty cycle calculation errors, thereby reducing 5th and 7111 harmonic components and improving torque ripple. Moreover, it ensures that even with 24 voltage vectors in action, the computational requirements remain low, thereby enhancing the efficiency of the algorithm execution.

Technical solution: To achieve the aforementioned objectives, the technical solution adopted by the present invention is as follows: a high-precision model predictive current control system for a dual three-phase permanent magnet motor, comprising hardware and software components; the hardware components include a dual three-phase permanent magnet motor, a direct current (DC) power supply, a pulse width modulation (PWM) module, an inverter, a position sensor, and a current sensor; and the software components include a synthesized 24 virtual voltage vector module, a speed controller, a coordinate transformation module, a delay module, a prediction module, a duty cycle calculation module, a simplification module, and a cost function module.

The dual three-phase permanent magnet motor is composed of two sets of three-phase winding with a phase shift of 30 degrees; the input terminal of the inverter is connected to the DC power supply, and the signal terminal of the inverter is connected to the PWM module; the inverter adopts a six-phase two-level topology, and an output terminal of the inverter is connected to A, B, C, D, E, F phases of the dual three-phase permanent magnet motor to convert PWM signals into a required six-phase sinusoidal AC current for driving the dual three-phase permanent magnet motor; the position sensor uses a rotary transformer, and the rotary transformer is connected to the dual three-phase permanent magnet motor; and the current sensor is connected to the inverter to sample the six-phase motor current.

The coordinate transformation module has an input terminal connected to the current sensor and an output terminal connected to the delay compensation module, to convert a six-phase current in a natural coordinate system into a current in a rotating coordinate system to achieve decoupling control the delay module has an input terminal connected to the coordinate transformation module and an output terminal connected to the prediction module, to compensate for a delay issue caused by digital system sampling.

The prediction module has an input terminal connected to the delay compensation module, the synthesized 24 virtual voltage vectors module, and the position sensor, to output a variation of a &q-axis current under an influence of different voltage vectors.

The speed controller is configured through P1 control to obtain a q-axis reference current, an error between a desired speed and an actual speed is input to an terminal of the speed controller,and an output terminal of the speed controller outputs a reference value of a q-axis current.

The duty cycle calculation module has an input terminal connected to the speed controller and the prediction module, to calculate optimal positions and duty cycles of the voltage vectors under different voltage vector effects.

The simplification module and the cost function module have input terminals connected to the duty cycle calculation module, to reduce a number of algorithm iterations and select an optimal vector and a duty cycle of the optimal vector; the PWM module has an input terminal connected to the cost function module, to convert the optimal vector and the duty cycle of optimal vector obtained by a software system into corresponding PWM signals, and the corresponding PWM signals are then outputted to the inverter to complete modulation and drive the dual three-phase permanent magnet motor to operate.

A control method of the high-precision model predictive current control system for the dual three-phase permanent magnet motor of the present invention comprises the following steps: Step 1: constructing 24 virtual voltage vectors.

Step 2: optimizing the switching sequence of the voltage vectors to achieve standardization.

Step 3: obtaining the motor speed, and position angle through the position sensor, and obtain the six-phase currents through the current sensor. Then, use the coordinate transformation module to obtain the currents in the rotating coordinate system.

Step 4: deriving the predictive model for the dual three-phase permanent magnet motor. Step 5: calculating the duty cycles for the voltage vectors using the method of minimum error.

Step 6: simplifying the process of iterating and searching for optimization.

Step 7: selecting the optimal voltage vector and the duty cycle of the optimal voltage vector based on the cost function, and outputs the duty cycle to the PWM module. The corresponding voltage vectors are generated through the inverter to complete the entire control process.

Further, the specific steps of Step 1 comprises: a dual three-phase permanent magnet motor is configured in a neutral point isolation manner and driven by a six-phase two-level voltage source inverter. Each bridge ann has two switch states since the complementary conduction state of the upper and lower switch devices in each bridge arm. The entire inverter has a total of 26 = 64 switch states. The 64 voltage vectors corresponding to the switching states are determined by the following equation: = -uck (s), +88a4 +s as -FS (2/±C a5±C Ci9) 3 = -Ud, (sA + sitea8 + s(a4 ±,S + sza+ spa9) Where, ct = ei30, SA-Si:represent the switch states of each bridge leg, u,,,B represents the voltage vector in the afl subspace, 71.9' represents the voltage vector in the xy subspace, /Id, represents the DC bus voltage. "1" indicates that the upper bridge leg is turned on, while "0" indicates that the upper bridge leg is turned off. The basic voltage vectors are numbered in the order of ABC and DEF, and the switch states combination is represented by converting these binary switch states to octal notation.

The principle of virtual voltage vector requires that the total components of the voltage vector on the harmonic subspace equal zero. The synthesis principle is as follows: 2=n1 =1 = 0 (2) = I 2=1 Where, my, and uy, represent the components of the fundamental voltage vector on the x-axis and y-axis, respectively. Dr represents the duty cycle of each fundamental voltage vector.

To ensure voltage utilization, the 12 largest vectors on the outer periphery of the fundamental subspace, along with one zero vector, are chosen as the basic voltage vectors for synthesizing the composite virtual voltage vectors. The synthesis follows the principle of adjacent triple vectors. The synthesis principle is as follows: Vi" yfi a a a 0 D3 (3) Ulst "1st 142nd "3,1 1 1 u20d " U2nd 3 'I 113 U2nd 3n-1 1 1 Where, V, represents the i-th synthesized virtual voltage vector, with i = 1, 2, 3.. 24. mist, u2,,d, and uird represent the first, second, and third basic voltage vectors, respectively. The superscripts "a", 'fl", "x", and "y" denote the components of the voltage vectors along the corresponding coordinate axes. DI, D2, D3, and Do represent the duty cycles of the first, second, third basic voltage vectors, and the zero vector, respectively.

It is stipulated that the magnitude of each basic voltage vector is 0.59/1k. with a starting position at 00 and an angular spacing of 150 between adjacent voltage vectors. Ultimately, 24 virtual voltage vectors are synthesized in the all subspace, with zero components in the xy subspace.

Further, the specific steps of Step 2 include: In order to ensure the synthesized virtual voltage vectors can be implemented in industrial applications, the switching sequence of the 24 virtual voltage vectors is optimized to achieve full standardization. At positions V2, VG, VIO, VI4, V18, and V22, an inner and outer-layer voltage vector synthesis method is used instead of the adjacent three-vector synthesis method. The final set of 24 virtual voltage vectors is shown in Table 1.

Tab.1 Distribution of 24 virtual voltage vectors.

V, Basic voltage vector The ratio of basic voltage vector V ul U^ 113 Uu Di 1): Di Do Vi 1155 1445 1144 110 0.034 0.443 0.477 0.046 V2 1144 /16S / /10 0.723 0.264 0 0.013 V3 1/44 1464 1/66 1/0 0.477 0.443 0.034 0.046 V4 1/44 1164 1/66 11, , 0.264 0.458 0.264 0.014 V-. 1144 1464 1166 tin 0.034 0.443 0.477 0.046 V6 1166 1124 / 110 0.723 0.264 0 0.013 V7 1466 1126 1422 110 0.477 0.443 0.034 0.046 Vs 1165 /426 1422 110 0.264 0.458 0.264 0.014 V9 1166 1426 1422 110 0.034 0.443 0.477 0.046 V16 1122 1136 / 110 0.723 0.264 0 0.013 V11 1122 /132 an an 0.477 0.443 0.034 0.046 V12 4/22 /432 4133 /40 0.264 0.458 0.264 0.014 V13 /122 1432 1133 110 0.034 0.443 0.477 0.046 V14 1133 1412 / U0 0.723 0.264 0 0.013 V19 1433 1413 1411 110 0.477 0.443 0.034 0.046 V16 1433 1113 1111 110 0.264 0.458 0.264 0.014 V17 1133 1413 UN U0 0.034 0.443 0.477 0.046 V18 111 1 1153 / U0 0.723 0.264 0 0.013 V19 Ul 1 1451 1155 110 0.477 0.443 0.034 0.046 V20 /411 1151 1155 U0 0.264 0.458 0.264 0.014 V21 1111 1451 U58. /to 0.034 0.443 0.477 0.046 V72 U55 1141 / /to 0.723 0.264 0 0.013 V23 1455 /445 1444 110 0.477 0.443 0.034 0.046 V24 ti55 1145 1444 an 0.264 0.458 0.264 0.014 Where, ui represents the corresponding basic voltage vector.

Further, the specific steps of Step 3 include: The position sensor measures the angular displacement and angular velocity of the rotor, which are then converted into electrical signals and transmitted to the controller. After decoding, the motor's speed and rotor position angle information are obtained.

The phase currents of the motor, namely 41, 49, Ic, ID, iE, and iF, are sampled by six current sensors. These variables are transformed from the natural coordinate system to the stationary coordinate system using the VSD coordinate transformation method. The transformation matrix for this process is as follows: 1 - 0 2 2 2 2 V-3 1_ 0 2 2 2 2 -1 e, 1 1 1 Nr 0 " 3 2 2 2 2 D (4) 101,r3 0 -1 o2 2 2 2 2 _i" _ 1 1 1 0 0 0 _0 0 0 1 1 1 _ Where, ea, rib:, ex, I y, toi, and /02 represent the currents in the stationary coordinate system along the a-axis, fl-axis, x-axis, y-axis, 01-axis, and 02-axis, respectively.

For the dual three-phase permanent magnet motor, only the fundamental components in the aft subspace participate in the electromechanical energy conversion.

To simplify the analysis, the stationary coordinate system is transformed into the synchronous rotating coordinate system, and the transformation matrix can be expressed as: id COO Silth cos8 0 0 0 0 -. (5) iq -siri9 o 0 () 0 0 a o2_ 0 0 1 0 0 0 0 1 0 0 0 0 1. 0 0 i ix iv 0 0 0 i ol 0 1_ i _ 02 Where, 6 represents the rotor position angle, and id and ig represent the currents in the d-axis and q-axis, respectively.

The current of the motor id,(k) in the dci rotating coordinate system is calculated by the aforementioned coordinate transformation module at time k.

Further, the specific steps of Step 4 include: In the model predictive control system based on virtual voltage vectors, the harmonic subspace can be neglected, thus only the relevant variables of the dual three-phase permanent magnet motor in the fundamental subspace need to be considered. By transforming variables into the rotating coordinate system, the voltage equation of the motor is obtained as follows: ud =Rsid + Ld=-CO e Lqicr (6) dt u = Lq-did +roeLdid+odp f dt Where, /Ea and lig represent the components of Us in the d-axis and q-axis, respectively. Rs is the stator resistance, La, Lg, id, and ig are the dq-axis inductance and currents, tiff is the magnitude of the permanent magnet flux, and co, is the electrical angular velocity.

By employing the Euler forward formula, equation (6) can be discretized as follows: { k (7) I d1 akl = id +Ts-Eted -RIdk +coeIgiqk] I Ld i lc' I = iqk +T *V -R ik -ro i k -a) y/ II L <I s.1 sq eI-ald 'e f q Where, the superscript "k" represents the real-time values of dq-axis current and voltage at time k, and the superscript "k+1" represents the predicted values of &q-axis current at time k+1. Ts denotes the control period.

To compensate for the delay drawback of the digital controller, a two-step prediction method is employed for delay compensation Equation (7) is predicted again to obtain the final prediction model as follows: re + [11,44' I I Rs idk I I + i kill! id (8) e <2, 1 k 111 +7;* [I, - Lciid4.11 -Ceelr I Lq Where, the superscript ''pre" represents the final predictive value of dq-axis current.

The cost function is defined as follows: J = (id - e)2 (9) Where, the superscript "*" represents the reference value of dq-axis current, and the C = 0 method is adopted.

Further, the specific steps of Step 5include: With reference to the clq coordinate system, note the predicted values of id and i, under zero voltage vector be point A (xi, yi), and the predicted values of id and i, under effective voltage vector be point B y2). The reference current is located at point C (xo, yo). The distance from point C to line AB represents the minimum value of the cost function and the point of minimum error. By finding the intersection point of the line perpendicular to AB passing through point C, the required voltage vector can be obtained, and thus the corresponding duty cycle can be determined.

The equation of the line AB: y x -x y, x2-x1 The equation of the line perpendicular to AB passing through point C is as follows: X1 X2 -X2 V - X ± V0 Xo 322 -11 12-Y1 The coordinates of the intersection point of the two lines can be obtained by solving equations (10) and (11) simultaneously: (10) xo(x, + (h -y9)2 +(x1 -)(21 -Y0)022 - (X1 -x-02 + -y2)2 -v, )2 + (x1 -x, )2 + (x, -x0)(x, -x, )(y, -y1) -x, )2 + 2c1duIv (12) Where, Cth", and 47,, represent the predicted values of the dq-aid s currents under the modified duty cycle. with

X0 = yo =14 = idit+1 +1; * I -1?,11±1 +0,.1414k-1 I Ld Yi=i "'A +1; *1-Rsigk-1 WeLdiark±1 ectifI Lq x2 =1)+1 Ts * I / idA -Rsidk coeLqi4A '1 I 1 Ld icik+1 +1; [uqk +1 R sig.), +1 coeLdidk +1 we clif /Lq The optimal duty cycle for the voltage vector can be determined based on the position of the intersection point.

* Pre,d0 (13) - du 4, -X2 Further, the specific steps of Step 6 include: The at? subspace is divided into four groups, namely GI, (52, (ii, and 04, using V4, VIO, VI6, and V22 as boundaries. The virtual voltage vectors contained in each group are shown in the following table: Tab.2 Virtual Voltage Vector Partitioning Rules NI1111 Voltage vectors Cil V22. V23, V21 VI. V2, V3, V1 G2 Vi V5. V6, V. Vs, V9. V10 G3 VIO. V11. V12, V13, V11, V15, V16 CT4 VI6. V17, V18 V19. V23. V31. V22 By substituting VI, V7, VI3, and V19 into equation (5), the values of the cost function for each voltage vector are obtained: J(V1), J(V7), J(Vii), J(V19). The voltage vector Vi, with the minimum cost function is selected, thus determining the optimal group.

Assuming that Via, determined in the above step is VI, the optimal region is GI.

Then, the cost function values for V23 and V3 in GI are calculated again. The voltage vector V2nd with the minimum cost function is selected, determining the second optimal group.

Assuming that V2nd determined in the above step is VI, the cost function values for VI and the adjacent voltage vectors, V24 and V2, are compared. The optimal cost function value is selected to determine the final voltage vector index Similarly, for other cases, the combinations of all groups are shown in Table 3. Tab.3 Combinations of all optimal voltage vector selections Group Vlsi V210t1 Group Vlsi Vaud Gi(V 0 V1 V2, V24 Ga(VI3) V11 V10, V12 Va V2, V4 VI a VI2, VI 4 V23 V22, V24 Vi5 V14, V16 02(V7) V5 V1, V6 04(VI9) V17 V16, VlS V7 Vi, Vs VI 9 VIR, V20 V9 VS, Via V21 V20, V22 After the aforementioned simplification process, the original prediction process, which required traversing 24 voltage vectors, only needs to traverse 8 vectors, reducing the computational burden of the algorithm and improving efficiency.

Further, the specific steps of Step 7 include: The 24 virtual voltage vectors are individually input into the prediction model. Through the simplification module, the optimal voltage vector and the duty cycle of the optimal voltage vector are selected. Then output those duty cycles to the PWM module for modulation, resulting in the generation of the corresponding voltage vector. This completes the entire control process.

The present invention has the beneficial effects of: 1) The present invention provides a high-precision model predictive current control system and control method for a dual three-phase permanent magnet motor. By increasing the number of virtual voltage vectors from 12 to 24, the modulation range is expanded without compromising voltage utilization efficiency.

2) The proposed method for solving the duty cycle takes into account both the c/-axis and q-axis currents, reducing the minimum error and obtaining the smallest value among all the cost functions, thus enhancing control accuracy.

3) Under the effect of the proposed duty cycle technique, the magnitude of each voltage vector can be flexibly adjusted, thereby improving control accuracy, reducing 5th and 7th harmonic components, and mitigating torque and flux ripples 4) The simplified vector selection method presented in the invention reduces the execution time of the model predictive control algorithm, improves algorithm efficiency, and can be extended to other multi-phase motor predictive control systems.

Description of the Drawings

FIG. 1 is the control principle of the method according to an embodiment of the present invention.

FIG. 2 is the topology of a six-phase voltage source inverter applying the method according to an embodiment of the present invention.

FIG. 3 is the spatial voltage vector diagram of the present invention: (a) afl subspace; (b) xy subspace.

FIG. 4 is the construction diagram of the maximum three-vector virtual voltage vectors designed by the present invention: (a) afi subspace; (b),cy subspace.

FIG. 5 is the switching sequence diagram designed by the present invention (a) before V2 correction; (b) after V2 correction.

FIG. 6 is the construction diagram of the inner and outer two-layer virtual voltage vectors designed by the present invention: (a) afi subspace; (b) Ay subspace.

FIG. 7 is the 24 virtual voltage vectors designed by the present invention FIG. 8 is the schematic diagram of the proposed method for calculating the minimum error duty cycle.

FIG. 9 is the schematic diagram of the simplified process designed by the present invention: (a) division of groups; (b) optimal region; (c) optimal voltage vectors.

FIG. 10 is the experimental waveform of the conventional 12 virtual voltage vectors without deadbeat duty cycle.

FIG. 11 is the experimental waveform of the present invention.

Detailed Description of the Embodiments

In order to further illustrate the objectives, technical solutions, and advantages of the present invention, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and should not be construed as limiting the scope of the present invention.

FIG. 1 illustrates the control framework of the present invention. The hardware components include a dual three-phase permanent magnet motor, a direct current (DC) power supply, a pulse width modulation (PWM) module, an inverter, a position sensor, and a current sensor; and the software components include a synthesized 24 virtual voltage vector module, a speed controller, a coordinate transformation module, a delay module, a prediction module, a duty cycle calculation module, a simplification module, and a cost function module.

The dual three-phase permanent magnet motor is composed of two sets of three-phase winding with a phase shift of 30 degrees. The input terminal of the inverter is connected to the DC power supply, and the signal terminal of the inverter is connected to the PWM module; the inverter adopts a six-phase two-level topology, and an output terminal of the inverter is connected to A, B, C, D, E, F phases of the dual three-phase permanent magnet motor to convert PWM signals into a required six-phase sinusoidal AC current for driving the dual three-phase permanent magnet motor. The position sensor uses a rotary transformer, and the rotary transformer is connected to the dual three-phase permanent magnet motor; and the current sensor is connected to the inverter to sample the six-phase motor current.

The coordinate transformation module has an input terminal connected to the current sensor and an output terminal connected to the delay compensation module, to convert a six-phase current in a natural coordinate system into a current in a rotating coordinate system to achieve decoupling control.

The delay module has an input terminal connected to the coordinate transformation module and an output terminal connected to the prediction module, to compensate for a delay issue caused by digital system sampling.

The prediction module has an input terminal connected to the delay compensation module, the synthesized 24 virtual voltage vector module, and the position sensor, to output a variation of a dq-axis current under an influence of different voltage vectors.

The speed controller is configured through PI control to obtain a q-axis reference current, an error between a desired speed and an actual speed is input to an terminal of the speed controller,and an output terminal of the speed controller outputs a reference value of a q-axis current.

The duty cycle calculation module has an input terminal connected to the speed controller and the prediction module, to calculate optimal positions and duty cycles of the voltage vectors under different voltage vector effects.

The simplification module and the cost function module have input terminals connected to the duty cycle calculation module, to reduce a number of algorithm iterations and select an optimal vector and a duty cycle of the optimal vector. The PWM module has an input terminal connected to the cost function module, to convert the optimal vector and the duty cycle of optimal vector obtained by a software system into corresponding PWM signals, and the corresponding PWM signals are then outputted to the inverter to complete modulation and drive the dual three-phase permanent magnet motor to operate. The implementation steps of the method mainly involve the following steps: Step I: constructing 24 virtual voltage vectors.

As shown in FIG. 2, the dual three-phase permanent magnet motor of the present invention is configured in a isolated neutral point method and driven by a six-phase two-level voltage source inverter. Due to the complementary conduction state of the upper and lower switches in each bridge leg, there are two switch states for each bridge leg, resulting in a total of 26=64 switch states for the entire inverter. The 64 voltage vectors corresponding to the switching states are determined by the following equation: + s a4 ± So a8 ± s a5 + s a9) (1) = 1U (s4 s Ba8 + sca4 + spa5 + sBa + s Ba9) 3 Where, a = e0, sA-57: represent the switch states of each bridge leg, miiirepresents the voltage vector in the a/3 subspace, 11.9. represents the voltage vector in the xy subspace, Ak, represents the DC bus voltage. "1" indicates that the upper bridge leg is turned on, while "0" indicates that the upper bridge leg is turned off. The basic voltage vectors are numbered in the order of ABC and DEF, and the switch states combination represented by converting these binary switch states to octal notation.

The resulting voltage vectors are shown in FIG. 3. FIG. 3(a) represents the voltage vectors in the czy3 subspace, which are responsible for participating in the electromechanical energy conversion. The 48 effective voltage vectors are divided into four layers, with increasing magnitudes from the innermost to the outermost: 0.173/1/dc, 0,333U, 0.4710,k, and 0.644Ak. FIG. 3 (b) represents the xy subspace, which is the harmonic subspace responsible for generating losses. Similarly, the magnitudes increase from the innermost to the outermost: 0.173 0.471lidc, and 0.644 tick.

In order to suppress the voltage vectors generated by the inverter in the harmonic subspace, the present invention proposes a novel method for synthesizing virtual voltage vectors. The principle of virtual voltage vector synthesis requires that the sum of the vectors' effects in the harmonic subspace is zero. The synthesis principles are as follows: =0 (2) Where, tt,, and try, represent the components of the fundamental voltage vector on the x-axis and y-axis, respectively. A represents the duty cycle of each fundamental voltage vector.

In order to ensure voltage utilization, the present invention selects twelve major vectors and one zero vector from the outermost layer of the fundamental subspace for synthesizing the virtual voltage vectors. The synthesis of a new virtual voltage vector control set is performed based on the principle of adjacent three vectors. As shown in FIG. 4, assuming the target voltage vector is V3. V3 forms an angle of 30°with the a-axis, and adjacent three voltage vectors of V3 are u66, um, and 1144. According to the principle of virtual voltage vector synthesis, the following equation can be deduced: Di-4+D, -4 +D, -=V -1136 +D2 -144 +D3 U413.4 = D, * tist,, +D, * u7,4 + D, * u4.4 =0 (3) -"))76 +D2 + D3 = V;:bj =0 1) +T) +1)3 + =1 Based on equation (3), Di, 1)2, D3, and Do can be determined.

Following the same principle, other virtual voltage vectors can be obtained accordingly. The synthesis principle can be summarized as follows: (4) Where, V, represents the i-th virtual voltage vector to be synthesized (i = 1, 2, 3... 24); uist, 142,,a, and Uhrd represent the first, second, and third basic voltage vectors, respectively. The superscripts "a", 'fl", "x", and "y" denote the components of the voltage vector along the corresponding coordinate axes. Di, D2, D3, and Do represent the duty cycles of the first, second, third basic voltage vectors, and zero vector, respectively.

Step 2: Optimize the switching sequence of the voltage vectors to achieve standardization.

FIG. 5(a) illustrates the switching sequence diagram of the virtual voltage vector V2 (angle of 150 with respect to the a-axis) synthesized using the adjacent three-vector principle. It can be observed from the figure that the switching sequence of phase F needs to be operated twice within one cycle, which is not favorable for the implementation of digital processors in industrial applications. To address this, a local adjustment is required for the specific position of vector V2. The method is as follows: as shown in FIG. 6, V2 is synthesized using an inner and outer layer, combining two voltage vectors in the same direction in the ap-subspace and opposite direction in the icy-subspace. The resulting switching sequence is depicted in FIG. 5(b), which meets industrial requirements. Similarly, at positions V2, V6, V10, V14, V18, and V22, the adjacent three-vector synthesis method is replaced with the inner and outer layer voltage vector synthesis method.

After applying the methods from the previous two steps, the final synthesis results in 24 standardized virtual voltage vectors, as shown in Table 1.

Tab.1 Distribution of 24 virtual voltage vectors.

V; Basic voltage vector The ratio of basic voltage vector V; u, Lb 113 Ilu DI 1)2 1)3 Do VI 1/55 1145 tun ito 0.034 0.443 0.477 0.046 V2 1144 /465 / fro 0.723 0.264 0 0.013 Vi 1144 1164 1166 UV 0.477 0.443 0.034 0.046 V4 1144 1164 1166 Hi i 0.264 0.458 0.264 0.014 Vs 1144 1164 1166 110 0.034 0.443 0.477 0.046 V6 1166 1424 / 110 0.723 0.264 0 0.013 V7 1166 1426 1122 110 0.477 0.443 0.034 0.046 VS 1166 1126 1122 an 0.264 0.458 0.264 0.014 VS //66 1116 1123 //6 0.034 0.443 0.477 0.046 V16 112, 1135 / U0 0.723 0.264 0 0.013 V11 /432 /433 tio 0.477 0.443 0.034 0.046 V12 11 1112 1131 110 0.264 0.458 0.264 0.014 -" Vli 1122 1432 1133 110 0.034 0.443 0.477 0.046 VI 4 1133 1412 / a,, 0.723 0.264 0 0.013 Vls 1133 1113 1111 /10 0.477 0.443 0.034 0.046 V16 1133 1413 1111 110 0.264 0.458 0.264 0.014 VI 7 1133 1113 Ill 1 110 0.034 0.443 0.477 0.046 Vis tin ti-H / 110 0.723 0.264 0 0.013 V111 1111 1111 1111 1/6 0.477 0.443 0.034 0.046 V20 1111 U51 U55 110 0.264 0.458 0.264 0.014 V21 ti I 1 1451 1155 tio 0.034 0.443 0.477 0.046 V22 1155 1141 / /to 0.723 0.264 0 0.013 V23 1/ s s 1145 1144 a,, 0.477 0.443 0.034 0.046 V24 U s s 1145 1144 1/0 0.264 0.458 0.264 0.014 The distribution of 24 virtual voltage vectors is illustrated in FIG. 7.

Step 3: The speed and angular position are obtained through the position sensor, while the six-phase currents are obtained through the current sensor. Subsequently, the coordinate transformation module is used to acquire the currents in the rotating coordinate system.

The position sensor measures the angular displacement and angular velocity of the rotor's shaft and converts the angular displacement and angular velocity into electrical signals transmitted to the controller. Upon decoding, the information of the motor speed and rotor angular position are obtained.

In this invention, six current sensors are used to sample the phase currents of the motor, denoted as iA, ID, lc, in, fr, and ir. The VSD coordinate transformation method is employed to convert the variables from the natural coordinate system to the stationary coordinate system, the transformation matrix is expressed as follows: IJy 3 1 -- 2 F3 2 0 1,. (5) ix 0 2..,U 2 1 -1 Jp.

lo I V7,3 2 1 2 0 I 2 __1 -2..,U VI 0 0 1 2 2 2 1 2 -sr3 1 1 V-3 2 2 2 2 1 0 0 I 0 1 1 Where, la, 1,8, tx, ty, to, and io2 represent the currents on u-axis, it-axis, x-axis, yams, oi-axis, and o2-axis in the stationary coordinate system, respectively.

For the dual three-phase permanent magnet motor, only the fundamental components in the aft subspace are involved in the electromechanical energy conversion. To facilitate simplified analysis, the stationary coordinate system is transformed into the synchronous rotating coordinate system. The transformation matrix is expressed as follows: id cos 0 sin 0 cos8 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 1_ fir (6) Ix -sin9 o 1/3 _io2 _ 0 o ix 0 0 1,, 0 iol _jo2 Where, 6 represents the rotor position angle, and id and I, represent the currents in the d-axis and q-axis, respectively.

The current of the motor id,(k) in the dci rotating coordinate system is calculated by the aforementioned coordinate transformation module at time k.

Step 4: Derivation of the predictive model for the dual three-phase permanent magnet motor.

The present invention is based on a model predictive control system using virtual voltage vectors, where the harmonic subspace can be neglected. Therefore, it is only necessary to consider the relevant variables of the dual three-phase permanent magnet motor in the fundamental subspace. By transforming control variables into the rotating coordinate system, the voltage equation of the motor is obtained as follows: di -u =Rsid + La= -ro di di (7) L +69 L i + (Del/If e Where, zid and zi, represent the components of Us in the d-axis and q-axis, respectively. R. is the stator resistance, La, Lq, id, and A, are the &q-axis inductance and currents, tiff is the magnitude of the permanent magnet flux, and we is the electrical angular velocity.

By employing the Euler forward formula, equation (7) can be discretized as follows: {i Pi = il +Ts -[u,k4 -1251i + oeLcii: ] I Ld i k 11 = i k -FT * [uk -R i k -Welididk -rimy HI Lc, Q Li, zi, Li Where, the superscript "k" represents the real-time values of dq-axis current and voltage at time k, and the superscript "k+1" represents the predicted values of &q-axis current at time k+1, i denotes the control period.

To compensate for the delay drawback of the digital controller, a two-step prediction method is employed for delay compensation Equation (8) is predicted again to obtain the final prediction model as follows: * (9) = cik +I +7; -Eu,41+1 -Rsisik+1 +oeLgic2k+1]1 Lci +7;*[14-1 -coeLdiak+1 -wit/ f]l Lq Where, the superscript "pre" represents the final predictive value of &q-axis current.

The cost function is defined as follows: qd* -ed/'")2 - (10) (8) Where, the superscript "*" represents the reference value of ckfraxis current, and the i = 0 method is adopted.

Step 5: Calculation of duty cycle for the voltage vector using the minimum error method.

FIG. 8 illustrates the calculation of the duty cycle using the minimum error approach. With the dq coordinate system as a reference, note the predicted values of id and under zero voltage vector be point A (xi, yi), and the predicted values of id and 1" under effective voltage vector be point B (x2, y2). The reference current is located at point C (ro, yo). The blue dashed line represents the range of voltage vector action under duty cycle adjustment.

According to the form of the cost function J, the distance between the predicted point (C, ) and the reference current point (1,* , ei) represents the value of the cost function. If duty cycle adjustment technique is not used, a complete voltage vector is applied within one control cycle, and the cost function is represented by the orange line. When the conventional (taxis current deadbeat duty cycle calculation method is adopted, the ordinate of the target point for the predicted value is i; . In this case, the intersection point of the parallel line to the id-axis passing through point C and line AB represents the predicted point for the deadbeat duty cycle method, and the corresponding cost function value is represented by the green line segment. Based on set theory, it is known that neither of these methods determines the duty cycle with the minimum value of the cost function. The point at the shortest distance from point C to line AB represents the minimum value of the cost function and the smallest error. Obtaining the intersection point of the line passing through point C and perpendicular to AB provides the desired voltage vector and subsequently determines the corresponding duty cycle.

The equation of the line AB:

Y X

( 1 1) X19 Yi X2 -The equation of the line perpendicular to AB passing through point C is as follows: y X1 X2 (12) ± yo.vo Y2 The coordinates of the intersection point of the two lines can be obtained by solving equations (10) and (11) simultaneously: _ xo(xi -x2)2 +x1(y1 -.Y2)2 +(xi -x2)(Y1-.Y0AY2 -30 (13) (y -v2)2 +yi.(xi. -x2)2 +(xi. -x0)(x1 -x2)(y, -y12) (xi -)2 +(Y -y2)2 Where, frth", and represent the predicted values of the dq-axis currents under the modified duty cycle. with *

A'11 = Id 'Yo

_

=tri +7. * [-R, k+1 (1),/,,Tiqk-11/ f".7 =q1 +Ts * [-R, iqk -1 - I/ ligeTj x.2.ruldr 11 Rs/ dk 1 ± 41,7 Ili Ld qk 11 T Fultir 11 R 41,711 vit 11 Lq The optimal duty cycle for the voltage vector can be determined based on the position of the intersection point.

do =1-di (14) Step 6: The process of traversal and optimization.

6.1: As shown in FIG. 9(a), the aft subspace is divided into four groups, named Gi, G2, G3, and Ga, with Va, Vi 0, V16, and V22 as boundaries. The table below illustrates the virtual voltage vectors contained in each group.

Tab.2 Virtual Voltage Vector Partitioning Rules Num Voltage vectors GI V22 * V23, V24 VI * V2 V3, V4 G2 Vi, Vo* Vs * Vg* Vs, Vs. V10 GO VIO* Vit. V12.* V13, VI I * Vii, Vii Go VIs* V17, VIS* Vii. V23. V2I* V22 6.2: By substituting V1, V7, V13, and V19 into equation (5), the values of the objective function for each voltage vector: J(Vi), 1(V2), 1(V13), 1(V19) can be obtained. The vector Visi with the minimum value of the cost function is selected, determining the optimal group.

6.3: As shown in FIG. 9(b), assuming that Vi st determined in the above step is VI, the optimal region is Gt. Then, the cost function values for V23 and V3 in G1 are calculated again. The voltage vector V2"d with the minimum cost function is selected, determining the second optimal group.

6.4: As shown in FIG. 9(c), assuming that Vznd determined in the above step is VI, compare the value of the cost function between Vi and adjacent voltage vectors of Vi, V24 and Vz. Select the optimal value of the cost function to determine the final index of the optimal voltage vector.

6.5: Similarly, for other cases, the combinations of all groups are shown in Table Tab.3 Combinations of all optimal voltage vector selections Group Vlst V2nd Group VI st V2t8i GI(VI) Vi V2, V21 G3(V13) Vii Vie, Viz V3 V2, V4 VI3 VI2, VI4 V23 V22, V24 V15 VI 4, VI 6 G2(V7) V5 V4, Vd G1(V19) VI5 Vld, V18 V7 V6, V8 Vie Vie, V20 Ve VS, VIO V2I V20, V22 After the aforementioned simplification process, the original prediction process, which required traversing 24 voltage vectors, only needs to traverse 8 vectors, reducing the computational burden of the algorithm and improving efficiency.

Step 7: The 24 virtual voltage vectors are individually input into the prediction model. Through the simplification module, the optimal voltage vector andduty cycle of the optimal voltage vector are selected. Then output the duty cycles to the PWM module for modulation, resulting in the generation of the corresponding voltage vector, to complete the entire control process.

FIG. 10 shows the experimental waveform of the conventional 12 virtual voltage vectors under the effect of deadbeat duty cycle technique, with a THD of 19.8% and torque ripple of 10.2 Nm. The current ripples of id, ig, ix, and iy are 0.86A, 0.35A, 0.89A, and 0.71A, respectively. FIG. 11 shows the experimental waveform under the proposed method in the present invention, with a THD of 7.5% and torque ripple of 5.29 Nm. The current ripples of id, icy, ix, and are 0.33A, 0.18A, 0.43A, and 0.46A, respectively. Compared to the conventional method, the present invention significantly improves the performance of the motor.

The above embodiments are provided for illustrating the design principles and features of the present invention. The purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The scope of protection of the present invention is not limited to the above embodiments. Therefore, any equivalent changes or modifications based on the principles and design ideas disclosed in the present invention are within the scope of protection of the present invention.

Claims (9)

1. Claims What is claimed is: 1. A high-precision model predictive current control system for a dual three-phase permanent magnet motor, comprising hardware and software components, wherein the hardware components comprise a dual three-phase permanent magnet motor, a direct current (DC) power supply, a pulse width modulation (PWM) module, an inverter, a position sensor, and a current sensor, and the software components comprise a synthesized 24 virtual voltage vector module, a speed controller, a coordinate transformation module, a delay module, a prediction module, a duty cycle calculation module, a simplification module, and a cost function module, the dual three-phase permanent magnet motor is composed of two sets of three-phase winding with a phase shift of 30 degrees, an input terminal of the inverter is connected to the DC power supply, and a signal terminal of the inverter is connected to the PWM module; the inverter adopts a six-phase two-level topology, and an output terminal of the inverter is connected to A, B, C, D, E, F phases of the dual three-phase permanent magnet motor to convert PWM signals into a required six-phase sinusoidal AC current for driving the dual three-phase permanent magnet motor; the position sensor uses a rotary transformer, and the rotary transformer is connected to the dual three-phase permanent magnet motor; and the current sensor is connected to the inverter to sample the six-phase motor current, the coordinate transformation module has an input terminal connected to the current sensor and an output terminal connected to the delay compensation module, to convert a six-phase current in a natural coordinate system into a current in a rotating coordinate system to achieve decoupling control; the delay module has an input terminal connected to the coordinate transformation module and an output terminal connected to the prediction module, to compensate for a delay issue caused by digital system sampling; the prediction module has an input terminal connected to the delay compensation module, the synthesized 24 virtual voltage vector module, and the position sensor, to output a variation of a &q-axis current under an influence of different voltage vectors; the speed controller is configured through PI control to obtain a q-axis reference current, an error between a desired speed and an actual speed is input to an terminal of the speed controller,and an output terminal of the speed controller outputs a reference value of a c/-axis current; the duty cycle calculation module has an input terminal connected to the speed controller and the prediction module, to calculate optimal positions and duty cycles of the voltage vectors under different voltage vector effects; and the simplification module and the cost function module have input terminals connected to the duty cycle calculation module, to reduce a number of algorithm iterations and select an optimal vector and a duty cycle of the optimal vector; the PWM module has an input terminal connected to the cost function module, to convert the optimal vector and the duty cycle of optimal vector obtained by a software system into corresponding PWM signals, and the corresponding PWM signals are then outputted to the inverter to complete modulation and drive the dual three-phase permanent magnet motor to operate.
2. 2. A control method of the high-precision model predictive current control system for the dual three-phase permanent magnet motor according to claim 1, characterized in that, the control method comprises the following steps: Step 1: constructing 24 virtual voltage vectors; Step 2: optimizing switching sequences of the voltage vectors to achieve standardization; Step 3: obtaining the motor speed, position angle through a position sensor, and obtaining the six-phase currents through a current sensor, followed by obtaining the currents in the rotating coordinate system through a coordinate transformation module; Step 4: deriving the predictive model for the dual three-phase permanent magnet motor; Step 5: calculating the duty cycle of the voltage vectors using the method of minimum error; Step 6: simplifying a process of traversal and optimization; Step 7: selecting the optimal voltage vector and duty cycles of optimal voltage vector through a cost function and outputting the duty cycles to the PWM module; modulating the corresponding voltage vectors through the inverter and completing the entire control process.
3. 3. The control method of the high-precision model predictive current control system for the dual three-phase permanent magnet motor according to claim 2, characterized in that, the specific steps of step 1 comprise: the dual three-phase permanent magnet motor is configured in a neutral point isolation manner and driven by a six-phase two-level voltage source inverter; each bridge arm has two switch states since the complementary conduction state of the upper and lower switch devices in each bridge arm; the entire inverter has a total of 26 = 64 switch states; and 64 voltage vectors corresponding to the switching states are determined by the following equation: 1U 8 5 9 (S4 ± SEC/ 4 ± S(7 e + spa+ sEa ± spa = - ) + SeCt +sp115 +sRa+ s Fa9) 3 " wherein, a = ei300, .5,4-.5F represent the switch states of each bridge leg, flap represents the voltage vector in the a/7 subspace, u,, represents the voltage vector in the xy subspace, lick represents the DC bus voltag; "1" indicates that the upper bridge leg is turned on, while "0" indicates that the upper bridge leg is turned off; The basic voltage vectors are numbered in the order of ABC and DEF, and the switch states combination represented by converting these binary switch states to octal notation; the principle of virtual voltage vector requires that the total components of voltage vector on the harmonic subspace equal zero; The synthesis principle is as follows: {E Dint = El Diur = 0 ID; =1 wherein, u,, and u,, represent the components of the fundamental voltage vector on the x-axis andy-axis, respectively; D, represents the duty cycle of each fundamental voltage (2) i=1 vector; to ensure voltage utilization, the 12 largest vectors on the outer periphery of the fundamental subspace, along with one zero vector, are chosen as the basic voltage vectors for synthesizing the composite virtual voltage vectors, the synthesis follows the principle of adjacent triple vectors; the synthesis principle is as follows: yx utr. Uj ufrd 0 D1 (3) vi) ulst 111211 nd n d 0 D2 111 1 143rd 1_ D, Do 1 1 wherein, V, represents the i-th synthesized virtual voltage vector, with i = 1, 2, 3... 24. Hist, mid, and Lori represent the first, second, and third basic voltage vectors, respectively; the superscripts "a", "fl", "x", and "y" denote the components of the voltage vectors along the corresponding coordinate axes; Di, 1)2, 1)3, and DO represent the duty cycles of the first, second, third basic voltage vectors, and the zero vector, respectively; and it is stipulated that the magnitude of each basic voltage vector is 0.59/1k. with a starting position at 00 and an angular spacing of 150 between adjacent voltage vectors; ultimately, 24 virtual voltage vectors are synthesized in the aft subspace, with zero components in the xy subspace.
4. 4. The control method of the high-precision model predictive current control system for the dual three-phase permanent magnet motor according to claim 2, characterized in that, the specific steps of step 2 comprise: in order to ensure the synthesized virtual voltage vectors can be implemented in industrial applications, the switching sequence of the 24 virtual voltage vectors is optimized to achieve full standardization; at positions V2, VG, VIO, V14, Vls, and V22, an inner and outer-layer voltage vector synthesis method is used instead of the adjacent three-vector synthesis method; the final set of 24 virtual voltage vectors is shown in Table 1; Tab.1 Distribution of 24 virtual voltage vectors.Basic voltage vector The ratio of basic voltage vector V, in 112 113 110 DI 1): 1)1 0: Vi 1455 1145 1444 110 0.034 0.443 0.477 0.046 V2 1144 146: / an 0.723 0.264 0 0.013 V: /144 /464 1166 /10 0.477 0.443 0.034 0.046 V4 /144 /46; /166 au 0.264 0.458 0.264 0.014 V5 1144 1164 1166 lb 0.034 0.443 0.477 0.046 V6 1116 1124 / an 0.723 0.264 0 0.013 V7 /166 /436 1117 Liu 0.477 0.443 0.034 0.046 Vs 1166 /425 1432 110 0.264 0.458 0.264 0.014 V9 1116 1426 1421 Uu 0.034 0.443 0.477 0.046 VII) 1122 1436 I UI) 0.723 0.264 0 0.013 VII 1122 1433 1133 110 0.477 0.443 0.034 0.046 V12 112, lir 1133 an 0.264 0.458 0.264 0.014 V13 II- 1432 1133 1,10 0.034 0.443 0.477 0.046 V14 1133 1413 / lb 0.723 0.264 0 0.013 V, s uss 1413 an 1 UO 0.477 0.443 0.034 0.046 V, 6 1433 1413 1111 140 0.264 0.458 0.264 0.014 V17 1/33 1413 1111 Un 0.034 0.443 0.477 0.046 Vls un 1153 I 110 0.723 0.264 0 0.013 VI9 11,1 141 1155 110 0.477 0.443 0.034 0.046 V29 U1 1 141 as: th, 0.264 0.458 0.264 0.014 V21 all 14:1 ass th, 0.034 0.443 0.477 0.046 V22 /133 /441 / au 0.723 0.264 0 0.013 V23 1155 1145 1144 110 0.477 0.443 0.034 0.046 V24 1499 1143 1144 Un 0.264 0.458 0.264 0.014 wherein, u; represents the corresponding basic voltage vector.
5. 5. The control method of the high-precision model predictive current control system for the dual three-phase permanent magnet motor according to claim 2, characterized in that, the specific steps of step 3 comprise: the position sensor measures the angular displacement and angular velocity of the rotor, which are then converted into electrical signals and transmitted to the controller; after decoding, the motor's speed and rotor position angle information are obtained; the phase currents of the motor, namely Li, iB, ic, in, iL, and L, are sampled by six current sensors; the sampled variables are transformed from the natural coordinate system to the stationary coordinate system using the VSD coordinate transformation method; the transformation matrix for this process is as follows: 101 1 1 0 1 0 2 2 11100 I 2 0 -1 0 -1 iii (4) 3 0 -I - 2 2 1 1 i 2 n 1 -2 \ 2 -2 An -- I A:3 2 0 -2 F 1 * An 2 1 0 1 wherein, i, 4i, ix, ty, 7,1, and /o2 represent the currents in the stationary coordinate system along the a-axis, fl-axis, x-axis, y-axis, oi-axis, and 02-axis, respectively; for the dual three-phase permanent magnet motor, only the fundamental components in the afi subspace participate in the electromechanical energy conversion; to simplify the analysis, the stationary coordinate system is transformed into the synchronous rotating coordinate system, and the transformation matrix can be expressed as: Ix cos& sin9 cos() 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 I 0 0 0 0 0 0 1_ a ( 5) /01 -sin() o " _ 0 o iol _jo2 _ 0 0 wherein, 0 represents the rotor position angle, and id and 1, represent the currents in the d-axis and q-axis, respectively; the current of the motor id,(k) in the dq rotating coordinate system is calculated by the aforementioned coordinate transformation module at time k.
6. 6. The control method of the high-precision model predictive current control system for the dual three-phase permanent magnet motor according to claim 2, characterized in that, the specific steps of step 4 comprise: in the model predictive control system based on virtual voltage vectors, the harmonic subspace can be neglected, thus only the relevant variables of the dual three-phase permanent magnet motor in the fundamental subspace need to be considered; by transforming variables into the rotating coordinate system, the voltage equation of the motor is obtained as follows: ud = did (6) Rsid + Ld esOeLqi dt u =R i +LdtweL4i + oecif f wherein, mg, and 15, represent the components of Us in the d-axis arid y-axis, respectively; Rs is the stator resistance, La, [ dq, id, and iq are the dy-axis inductance and currents, wais the magnitude of the permanent magnet flux, and co, is the electrical angular velocity; by employing the Euler forward formula, equation (6) can be di scretized as follows: = + [te,k4 + o Ai]," ] I Ld qk 11 =1,7k *Iu _Rsiqk -tveL di dk -12mv ([1 lig wherein, the superscript "k" represents the real-time values of dq-axis current and voltage at time k, and the superscript "k+1" represents the predicted values of dy-axis current at time k+ Ts denotes the control period; to compensate for the delay drawback of the digital controller, a two-step prediction method is employed for delay compensation. Equation (7) is predicted again to obtain the final prediction model as follows: {idr =1dA -1 -FT *[uh-k1 -Rs/dk-sl ± W L A -Ad (8) s d e q1 ql] I L Pre 1 1 + = k 1 -I 1-T -[u Rk 1 -i -coef,dici coew f] I I,q 11 s 17 s 11 wherein, the superscript "pre" represents the final predictive value of dq-axis current, the cost function is defined as follows: J = (z - -)2 pre) 2 (9) wherein, the superscript "*" represents the reference value of dy-axis current, and the.1 c*, = 0 method is adopted.
7. 7. The control method of the high-precision model predictive current control system for the dual three-phase permanent magnet motor according to claim 2, characterized in that, the specific steps of step 5 comprise: with reference to the dy coordinate system, note the predicted values of id and iq under zero voltage vector be point A (11, yi), and the predicted values of id and ig under effective voltage vector be point B (12, y2); the reference current is located at point C (xo, yo); the distance from point C to the line AB represents the minimum value of the (7) cost function and the point of minimum error; by finding the intersection point of the line perpendicular to AB passing through point C, the required voltage vector can be obtained, and thus the corresponding duty cycle can be determined; the equation of the line AB: y - x - (10) x2 -the equation of the line perpendicular to AB passing through point C is as follows: -X2 x1-x2 Y = x + yo xo Y? -Y1 the coordinates of the intersection point of the two lines can be obtained by solving equations (10) and (11) simultaneously: (x1 -X2)2 ± (yi -)2)2 ± x2)021 -Y0XY2 Y1) (Yi -) 2 + Y1 (XI -x2)2 + 071 -X0 XXI -x,)(, Yi) -x2)2 +(y1 -)2 (12) (xi -x2)2 kk (Y1 32)2 wherein, C. and ti";,", represent the predicted values of the dq-axis currents under modified duty cycle; with ro = , Yo = x-1 = 11-1 +7; *1-Rsik+1 //gigk-1 / yi = +7; -CO, Ldidk+1 -0), r / Lq Jr 2 fir + .116 11 Rsidk 1 +U)eLq!qkll]ILd v, /clic 1 + Ts pilcir 11 R 41(711 ruelididk roe wit I / Lci the optimal duty cycle for the voltage vector can be determined based on the position of the intersection point.ddun d 0 -I d, (13) 11 - 8. The control method of the high-precision model predictive current control system for the dual three-phase permanent magnet motor according to claim 2, characterized in that, the specific steps of step 6 comprise: the aft subspace is divided into four groups, namely Gi, G2, G3, and G4, using Va, V16, and V22 as boundaries; the virtual voltage vectors contained in each group are shown in the following table: Tab.2 Virtual Voltage Vector Partitioning Rules Num Voltage vectors GI V22. V23, V21. Vi, V2 V3, V1 Ci2 Vi, V5. Vs, V7. Vs, V9. V10 V10. VII. Viz, VG, Vii VG, VG G4 V169 VI 7, V1R, V199 V20, V219 V22 by substituting V1, V7, V13, and V19 into equation (5), the values of the cost function for each voltage vector are obtained: J(Vt), J(V7), J(V13), J(V19); the voltage vector Vist with the minimum cost function is selected, thus determining the optimal group; assuming that Visi determined in the above step is Vi, the optimal region is GI; then, the cost function values for V23 and V3 in GI are calculated again; the voltage vector V24 with the minimum cost function is selected, determining the second optimal group; assuming that V2nd determined in the above step is VI, the cost function values for VI and the adjacent voltage vectors, V24 and V2, are compared; the optimal cost function value is selected to determine the final voltage vector index; similarly, for other cases, the combinations of all groups are shown in Table 3: Tab.3 Combinations of all optimal voltage vector selections.Group VIsi V2nd Group Yrs/ Vtur.GI(VI) VI V2, V21 GAVIS) VII Vie, VI2.

   Vs Vs, V4 Vi
8. S V12, V14 V23 V22, V24 VI5 Vii, Vi6 (32(V7) V5 V1, Vs G1(V19) VI' V16, V18 V7 Vs, Vs Viii V18, V20 VO V. Vii V21 V20, V22 after the aforementioned simplification process, the original prediction process, which required traversing 24 voltage vectors, only needs to traverse 8 vectors, reducing the computational burden of the algorithm and improving efficiency.
9. 9 The control method of the high-precision model predictive current control system for the dual three-phase permanent magnet motor according to claim 2, characterized in that, the specific steps of step 7 comprise: the 24 virtual voltage vectors are individually input into the prediction model; through the simplification module, the optimal voltage vector an corresponding duty cycle of the optimal voltage vector are selected; then output those duty cycles to the PWM module for modulation, resulting in the generation of the corresponding voltage vector, to complet the entire control process.