A Model Predictive Current Control System and Method for a Dual Three-Phase Permanent Magnet Motor

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

Description

The present invention belongs to the field of predictive control technologies for multiphase motors, and particularly relates to a 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 Publication No.: CN 114142784 B (Patent Application 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 Publication No.: CN 114785226 A (Patent Application 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 7th 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 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 t / ^-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 g-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 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 I 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 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. The 64 voltage vectors corresponding to the switching states are determined by the following equation: uap = yUdcUA +scai +sDa + sEa5 +spa9) [ (1) =—Udc{sA +sBc? +sca^ +sDa5 +sEa + sFa9) Where, a = e'’(l. sa~$f represent the switch states of each bridge leg, represents the voltage vector in the a / i subspace, uxy represents the voltage vector in the xy subspace, Ude 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: Where, uxi and uyi represent the components of the fundamental voltage vector on the x-axis and j-axis, respectively. D, 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: v“ T i vf c «11 «ld “2nd ua “3rd UP “3rd 0 0 d2 Y* = «1 U2nd U3rd 0 P v? «11 U2nd U3rd 0 Do_ 1 1 1 1 1 (3) Where, V, represents the / -th synthesized virtual voltage vector, with i = 1, 2, 3 ... 24. Hist, H2nd, and H3rd represent the first, second, and third basic voltage vectors, respectively. The superscripts "a", "P", "x", and "j" denote the components of the voltage vectors along the corresponding coordinate axes. Di, Di, and £>o 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 0.59¾ with a starting position at 0° and an angular spacing of 15° between adjacent voltage vectors. Ultimately, 24 virtual voltage vectors are synthesized in the up 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, Ve, V10, V14, Vis, 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. «i «2 U «0 Di n2 d3 Do V1 «55 «45 U44 «0 0.034 0.443 0.477 0.046 v2 «44 «65 I «0 0.723 0.264 0 0.013 v3 «44 «64 «66 «0 0.477 0.443 0.034 0.046 V4 «44 «64 «66 «0 0.264 0.458 0.264 0.014 v5 «44 «64 «66 «0 0.034 0.443 0.477 0.046 v6 #66 «24 / «0 0.723 0.264 0 0.013 V7 «66 «26 «22 «0 0.477 0.443 0.034 0.046 V8 «66 «26 «22 «0 0.264 0.458 0.264 0.014 v9 «66 «26 «22 «0 0.034 0.443 0.477 0.046 V10 «22 «36 / «0 0.723 0.264 0 0.013 V11 «22 «32 «33 «0 0.477 0.443 0.034 0.046 V12 «22 «32 »33 «0 0.264 0.458 0.264 0.014 V13 «22 «32 «33 «0 0.034 0.443 0.477 0.046 V14 «33 «12 / «0 0.723 0.264 0 0.013 V15 «33 «13 «11 «0 0.477 0.443 0.034 0.046 V16 «33 «13 «11 «0 0.264 0.458 0.264 0.014 V17 «33 «13 «11 «0 0.034 0.443 0.477 0.046 V1S un «53 / «0 0.723 0.264 0 0.013 V19 Un «51 «55 «0 0.477 0.443 0.034 0.046 V20 un «51 «55 «0 0.264 0.458 0.264 0.014 V21 un «51 «55 «0 0.034 0.443 0.477 0.046 v22 U55 «41 / «0 0.723 0.264 0 0.013 V23 Un «45 «44 «0 0.477 0.443 0.034 0.046 V24 U55 «45 «44 «0 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 11 11 24 rotor, which are then converted into electrical signals and transmitted to the controller. 5 After decoding, the motor's speed and rotor position angle information are obtained. The phase currents of the motor, namely ia, ib, ic, Id, Je, 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: Where, ia, ip, ix, iy, ioi, and ioi represent the currents in the stationary coordinate system along the a-axis, / / -axis, x-axis, j-axis, oi-axis, and 02-axis, respectively. For the dual three-phase permanent magnet motor, only the fundamental components in the a / 3 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: COS 0 sin0 0 0 0 0 ia ‘q Sill 0 COS0 0 0 0 0 ix 0 0 1 0 0 0 z* 0 0 0 0 1 0 0 ’y U 0 0 0 0 1 0 u jo2 _ 0 0 0 0 0 1 jo2_ (5) Where, 0 represents the rotor position angle, and id and iq represent the currents in the J-axis and < / -axis, respectively. The current of the motor idq(k) in the dq 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: to — R..ij 4-Lj--a^Ld,, « dz q q (b (6) Uq = Rdq aeLd’d + Where, Ud and uq represent the components of Us in the t / -axis and g-axis, respectively. Rs is the stator resistance, Ld, Lq, id, and iq are the c / g-axis inductance and currents, q / t is the magnitude of the permanent magnet flux, and coe is the electrical angular velocity. By employing the Euler forward formula, equation (6) can be discretized as follows: ^k+\ = ik +T^ ^R^k Where, the superscript "k" represents the real-time values of t / g-axis current and voltage at time k, and the superscript "k+l" represents the predicted values of db-axis current at time k+l. 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: • pre • . r.,k+l d 4 £+1 . t • k+l-i j t ld ~d +^s'lUd ~^sld + (De^qlq V ^d ‘ (o) lq ~lq +^s’lUq ~^slq ~ ae^dld ~ fd Lq Where, the superscript "pre" represents the final predictive value of d^-axis current. The cost function is defined as follows: j^qd-id^^+^q-iq^f (9) Where, the superscript represents the reference value of d^-axis current, and the C = 0 method is adopted. Further, the specific steps of Step 5 include: With reference to the dq coordinate system, note the predicted values of id and iq under zero voltage vector be point A (xi, ji), and the predicted values of id and iq under effective voltage vector be point B (x2, yd- The reference current is located at point C (xo, Jo). 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: (io) y2-vi x2-xj The equation of the line perpendicular to AB passing through point C is as follows: The coordinates of the intersection point of the two lines can be obtained by solving equations (10) and (11) simultaneously: ^(¾ -¾)2 + xi Qi -½)2 +(¾ - v )(¾ -ToXt? -¾) (¾-¾) +(¾-½) (¾ — V) + (½ -½) (12) Where, and fqyuty represent the predicted values of the <4 / -axis currents under the modified duty cycle, with xo =C -y^C *1 =' / +1 +TS-[~Rsidk+1 + Pi 5 A =iqk+l +TS-[^Rsiq^ ^(O^1 ^aWf\!Lq x2 =i / +l +TS X+1 ^Rsidk+1 +(oJ^iqM]ILd =<^} +K X' ~Rsik+i-(oeLdidM ~^f]!Lq The optimal duty cycle for the voltage vector can be determined based on the 11 11 24 position of the intersection point. (13) Further, the specific steps of Step 6 include: 10 The aft subspace is divided into four groups, namely Gi, G2, G3, and G4, using V4, V10, 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 Num Voltage vectors Gi V22, V23, V24, Vi, V-. Vh V4 G2 V>. Vs, Ve, V7, Vs, Vs, Vio Ga Vio, Vn. V12, Via, V14, V15, Vis G4 Vis, V17, Vis, Via, V20, V21, V22 By substituting Vi, V7, V13, and V19 into equation (5), the values of the cost 15 function for each voltage vector are obtained: / (Vi), J(Vis), J(Vi9). The voltage vector Vur with the minimum cost function is selected, thus determining the optimal group. Assuming that Vi,7 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 \ 2nd with the minimum cost function is selected, determining the second optimal group. Assuming that Nind 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 Vw Vw Group Vu( Vw Gi(Vi) V1 V2,V24 G3(V13) Vn Vio, V12 V3 V V4 V13 V12, V14 V23 V22. V24 V15 V14, V16 G2(V7) V5 V4,Vs G4(V19) V17 V16, V18 V7 V6,Vs V19 V18, V20 v9 Vs, V10 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 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 d-axis and < / -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) ap subspace; (b) xy subspace. FIG. 4 is the construction diagram of the maximum three-vector virtual voltage vectors designed by the present invention: (a) ap subspace; (b) xy 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) op subspace; (b) xy 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 di / -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 g-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 1: 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: 1 ua(} =-Udc(sA +¾0 +¾0 + ¾^¾° + sFa ) (1) 1 8 4 5 9, =—Udc(sA+sBa +sca + sDa +sEa+sFa ) Where, a = e|3°, $a~$f represent the switch states of each bridge leg, represents the voltage vector in the subspace, uxy represents the voltage vector in the xy subspace, Ude 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 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 Z 4a , 0.333Ude, QAllUdc, and 0.644½. 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: 0.1734^, 0.333Ude, QAlXUdc, and 0.644½. 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: i—1 i—1 fry X . / =1 Where, uxt and uyt represent the components of the fundamental voltage vector on the x-axis and j-axis, respectively. Dt 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 «66, «64, and «44. According to the principle of virtual voltage vector synthesis, the following equation can be deduced: A '«66 . Fi tia ±n „a — \7a 'u64 *W44 - yobj D -iD u\ u66 +D2-«f4+£)3-4=V^. D +D2nd + D3 -u^ = N*b] = 0 ’ M66 + D: > + D3 + DQ = 1 (3) 5 Based on equation (3), A, £>2, Di, 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: 11 11 24 ’ i vf A, X' U2nd ULd U3rd U3rd 0 0 d2 Y* = c U2nd U3rd 0 W c ^2nd uy u3rd 0 Do 1 1 1 1 1 (4) Where, V, represents the z-th virtual voltage vector to be synthesized (z = 1, 2, 3... 10 24); uist, uind, and u^rd represent the first, second, and third basic voltage vectors, respectively. The superscripts "a", " / T, "x", and "y denote the components of the voltage vector along the corresponding coordinate axes. Di, Di, Di, and Do represent the duty cycles of the first, second, third basic voltage vectors, and zero vector, respectively. 15 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 15° 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 20 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 a / >-sub space and opposite direction in the xy-subspace. The resulting switching sequence is depicted in FIG. 5(b), which meets industrial requirements. Similarly, at positions V2, ^6, V10, V14, Vis, and V22, the adjacent three-vector synthesis method is replaced with the inner and outer layer voltage 5 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. 11 11 24 Tab.l Distribution of 24 virtual voltage vectors. V / Basic voltage vector The ratio of basic voltage vector V, «i «2 »3 «0 Di d2 D, Do V1 «55 «45 «44 «0 0.034 0.443 QA11 0.046 V2 «44 «65 / «0 0.723 0.264 0 0.013 V3 «44 «64 W66 »0 0.477 0.443 0.034 0.046 V4 «44 «64 «66 «0 0.264 0.458 0.264 0.014 V5 «44 «64 «66 »0 0.034 0.443 0.477 0.046 v6 «66 «24 / «0 0.723 0.264 0 0.013 v7 «66 «26 »22 «0 0.477 0.443 0.034 0.046 v8 «66 «26 «22 »0 0.264 0.458 0.264 0.014 V9 «66 «26 «22 «0 0.034 0.443 0.477 0.046 V10 «22 «36 / «0 0.723 0.264 0 0.013 V11 «22 «32 «33 «0 0.477 0.443 0.034 0.046 V12 «22 «32 «33 «0 0.264 0.458 0.264 0.014 v13 «22 «32 «33 »0 0.034 0.443 0.477 0.046 V14 «33 «12 / «0 0.723 0.264 0 0.013 V15 «33 «13 «11 »0 0.477 0.443 0.034 0.046 v16 «33 «13 »11 «0 0.264 0.458 0.264 0.014 V17 «33 «13 «11 »0 0.034 0.443 0.477 0.046 Vis «11 «53 / »0 0.723 0.264 0 0.013 V19 «11 «51 »55 »0 0.477 0.443 0.034 0.046 v20 «11 «51 »55 «0 0.264 0.458 0.264 0.014 V21 «11 «51 »55 »0 0.034 0.443 0.477 0.046 V22 «55 «41 / »0 0.723 0.264 0 0.013 V23 «55 «45 »44 «0 0.477 0.443 0.034 0.046 V24 »55 «45 »44 «0 0.264 0.458 0.264 0.014 The distribution of 24 virtual voltage vectors is illustrated in FIG. 7. 10 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 15 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 1a, ib, ic, h, ie, and if. 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: 1 _1 2 _1 2 > / 3 V V 0 5 / 3 JI 0 u p 2 2 2 iB ix 1 1 1 1 -Ji i ............... . ............. . ............................ ............................ 0 2 2 2 2 q] x / 3 > / 3 1 1 iE 0 __ — — -1 o2 J 2 2 2 2 Jr. 1 1 1 0 0 0 0 0 0 1 1 1 Where, ia, ip, ix, iy, i;\, and i02 represent the currents on a-axis, / / -axis, x-axis, y-axis, oi-axis, and 02-axis in the stationary coordinate system, respectively. For the dual three-phase permanent magnet motor, only the fundamental components in the afi 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# -sin# ix 0 ‘y 0 ioi 0 io2~ 0 sin# cos# 0 0 0 0 0 0 o' 0 la J 0 1 0 0 0 ix 0 0 1 0 0 ly 0 0 0 1 0 ioi 0 0 0 0 1 (6) Where, 0 represents the rotor position angle, and id and iq represent the currents in the t / -axis and g-axis, respectively. The current of the motor idq(k) in the dq 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: Ud = RSid^Ld^—(»eLqiq CiMy ZM — 4- I- G)j 4- CD Al / / q s $ Q e a a er j Where, Ud and uq represent the components of Us in the d-axis and g-axis, respectively. Rs is the stator resistance, Ld, Lq, id, and iq are the dg-axis inductance and currents, y / f is the magnitude of the permanent magnet flux, and coe is the electrical angular velocity. By employing the Euler forward formula, equation (7) can be discretized as follows: 'A+1 =A +TA*kd~IUdk +^eLqiAILd ‘ (o) iq+" = ~RA^AdA Where, the superscript "k" represents the real-time values of dg-axis current and voltage at time k, and the superscript "Ml" represents the predicted values of dg-axis current at time Ml. 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 (8) is predicted again to obtain the final prediction model as follows: " ■ pre _ ■ Ml T r„Ml _ p : Ml r - Ml, , j ld ld lud ^sld '^e^q'q i' ^d * ^pre ,[WM1 Where, the superscript "pre" represents the final predictive value of dg-axis current. The cost function is defined as follows: J = Qd-idpriY+(.^iqpre')2 (10) Where, the superscript "*" represents the reference value of dg-axis current, and the C = 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 iq under zero voltage vector be point A (xi, ji), and the predicted values of id and iq under effective voltage vector be point B (x2, J2). The reference current is located at point C (xo, 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 (fr , f” ) and the reference current point (id, i*) 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 < / -axis current deadbeat duty cycle calculation method is adopted, the ordinate of the target point for the predicted value is / *. 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: = (11) X2-*l The equation of the line perpendicular to AB passing through point C is as follows: x, -x7 x, x, v = —5----x + y0—4 -x0 (12) T2 ^1 yi The coordinates of the intersection point of the two lines can be obtained by solving equations (10) and (11) simultaneously: (Xj — x2) + Cq — ) ToGi -yif+ydy -¾)2+(¾ -^)(¾ -^)(½ -¾) (¾^¾)2+(¾^+2)2 (13) Where, f^ty and represent the predicted values of the t / ^-axis currents under the modified duty cycle. 5 with +<> ’Vo = _ ■ A-+1 rp tn j k+1 T • A-+11 / T •*1 ld 1 s L 1Ysld ' ^e^q1 q J1 ^d yi =i^ +TS-[-Rsiqk+l] ~^f] / Lq *2 = / / +1 +TS -[^+1 ~Rsidk+1 +^eL9iqk+1] / Ld y2 =iqk+1 +TS X' -Ryqk+1 -'d-d' ~WfVLq The optimal duty cycle for the voltage vector can be determined based on the position of the intersection point. = A1 d _duty ¾ — x2 (14) 10 Step 6: The process of traversal and optimization. 6.1: As shown in FIG. 9(a), the afi subspace is divided into four groups, named Gi, G2, G3, and G4, with V4, Vw, Vi6, 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, V . V> V3, Vi G2 Vs, v5. Ve, V7, Vs, Vs, V10 G3 V10. Vu, V12, V13, Vi,i, V15. Vis G4 Vie, V17, Vis, Vis, V20, V21, V22 15 6.2: By substituting Vi, V7, V13, and V19 into equation (5), the values of the objective function for each voltage vector: J(Vi), Jiyd), J(Vi9) can be obtained. The vector Vm 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^ 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 Vanj with the minimum cost function is selected, determining the second optimal group. 5 6.4: As shown in FIG. 9(c), assuming that ¥2«? determined in the above step is ¥1, compare the value of the cost function between ¥1 and adjacent voltage vectors of ¥1, V24 and ¥2. 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 10 3. Tab.3 Combinations of all optimal voltage vector selections. 11 11 24 Group Vu, V'2nd Group Vu, Nlnd Gi(Vi) Vi V2,V24 G3(Vb) V11 V10, V12 V3 V2,V4 V13 V12, V14 V23 V22. V24 V15 V14, V16 G2(V7) V5 V4, Vo G4(V19) V17 V16, V18 V7 Ve,Vs V19 V18, V2o V9 V8, V10 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. 15 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. 20 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, iq, ix, and iy are 0.86A, 0.35A, 0.89 A, and 0.71 A, 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, iq, ix, and iy are 0.33 A, 0.18A, 0.43 A, 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 5 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 defined by the appended Claims. 11 11 24

Claims

What is claimed is:

1. A 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, tooutput a variation of a t / ^-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 g-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. A control method of the 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. The control method of the 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:(1)wherein, a = e|3°, sa~sf represent the switch states of each bridge leg, uap represents the voltage vector in the aft subspace, uxy represents the voltage vector in the xy subspace, Ude 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: / =1 / =1n1°.-'(2)wherein, uXi and uyi represent the components of the fundamental voltage vector on the x-axis and y-axis, respectively; D, represents the duty cycle of each fundamental voltagevector;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:v« ua u2nd U3rd 0 V / U2nd ^3rd 0 D1 = UXSt U2nd U3rd 0 D3 (3) “L U2nd U3rd 0 1 1 1 1 1wherein, V / represents the z-th synthesized virtual voltage vector, with i = 1, 2, 3 ...

24. «ist, «2nd, and «3rd represent the first, second, and third basic voltage vectors, respectively; the superscripts "a", "x", and "j" denote the components of the voltagevectors along the corresponding coordinate axes; Di, Dz, Di, and Do represent the duty cycles of the first, second, third basic voltage vectors, and the zero vector, respectively; andit is stipulated that the magnitude of each basic voltage vector is 0.59Udc, with a starting position at 0° and an angular spacing of 15° between adjacent voltage vectors; ultimately, 24 virtual voltage vectors are synthesized in the subspace, with zero components in the xy subspace.

4. The control method of the 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, Ve, V10, V14, Vis, 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 vector11 11 24V, «1 «2 «3 «0 Di d- Di Do V1 «55 «45 «44 «0 0.034 0.443 0.477 0.046 v2 «44 «65 / «0 0.723 0.264 0 0.013 v3 «44 «64 «66 «0 0.477 0.443 0.034 0.046 v4 «44 «64 «66 «0 0.264 0.458 0.264 0.014 v5 «44 «64 «66 «0 0.034 0.443 0.477 0.046 v6 «66 «24 / «0 0.723 0.264 0 0.013 v7 «66 «26 «22 «0 0.477 0.443 0.034 0.046 v8 «66 «26 «22 «0 0.264 0.458 0.264 0.014 Vs «66 «26 «22 «0 0.034 0.443 0.477 0.046 Vio «22 «36 / «0 0.723 0.264 0 0.013 Vn «22 «32 «33 «0 0.477 0.443 0.034 0.046 V12 «22 «32 «33 «0 0.264 0.458 0.264 0.014 V13 «22 «32 «33 «0 0.034 0.443 0.477 0.046 V14 «33 «12 / «0 0.723 0.264 0 0.013 V15 «33 «13 «11 «0 0.477 0.443 0.034 0.046 V16 «33 «13 «11 «0 0.264 0.458 0.264 0.014 V17 «33 «13 «11 «0 0.034 0.443 0.477 0.046 Vis «11 «53 / «0 0.723 0.264 0 0.013 V19 «11 «51 «55 «0 0.477 0.443 0.034 0.046 V20 «11 «51 «55 «0 0.264 0.458 0.264 0.014 V21 «11 «51 «55 «0 0.034 0.443 0.477 0.046 V22 «55 «41 / «0 0.723 0.264 0 0.013 V23 «55 «45 »44 «0 0.477 0.443 0.034 0.046 v24 «55 «45 «44 «0 0.264 0.458 0.264 0.014wherein, Hi represents the corresponding basic voltage vector.

5. The control method of the model predictive current control system for the dual three-phase permanent magnet motor according to claim 2, characterized in that, thespecific steps of step 3 comprise:5 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 u, ib, ic, id, Ie, and if, are sampled by six current sensors; the sampled variables are transformed from the natural coordinate system to 10 the stationary coordinate system using the VSD coordinate transformation method; the transformation matrix for this process is as follows:2> / 3 1'3(4)wherein, ia, ip, ix, iy, ioi, and iO2 represent the currents in the stationary coordinate system along the a-axis, ^-axis, x-axis, y-axis, m-axis, and 02-axis, respectively;for the dual three-phase permanent magnet motor, only the fundamental components in the 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:COS 0 sin# 0 0 0 0 ia Sill 0 cos# 0 0 0 0 ix 0 0 1 0 0 0 'y 0 0 0 1 0 0 ly ioi 0 0 0 0 1 0 ioi jo2 _ 0 0 0 0 0 1 jo2_(5)wherein, 0 represents the rotor position angle, and id and ig represent the currents in the d-axis and g-axis, respectively;the current of the motor idq(k) in the dq rotating coordinate system is calculated by the aforementioned coordinate transformation module at time k.

6. The control method of the 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:uddruq =Rsiq +Lq^+(oeLdid + aeVf(6)wherein, Ud and uq represent the components of Us in the <7-axis and g-axis, respectively; Rs is the stator resistance, Ld, Lq, id, and iq are the r / g-axis inductance and currents, y / yis the magnitude of the permanent magnet flux, and coe is the electrical angular velocity; by employing the Euler forward formula, equation (6) can be discretized as follows:• £+1ld■ £+1= hk+Ts-l«d-Rshk+^Lqiqk] / Ld— i „ + 7L * Rj. cd^L ,iico 31 / / 1 / L.^if s *- if s if e a a id * j if(7)wherein, the superscript "k" represents the real-time values of dg-axis current and voltage at time k, and the superscript "£+1" represents the predicted values of r / g-axis current at time £+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:• pre • £+1 . r.,k+l o • £+1 . t • k+\ i / rld ~ld +Rs'lud ~Rsld + ®3qlq U U• pre • £+1 . n-i r £+1 o • £+1 ,, t • k+\ ,, ,,, -] / tlq ~lq +Rs'luq ~Rslq ~^eRdld ~a,eV / f}l Lq(8)wherein, the superscript "pre" represents the final predictive value of dry-axis current;the cost function is defined as follows:-id^f +(iq-iqpKY(9)wherein, the superscript represents the reference value of dry-axis current, and the i d = 0 method is adopted.

7. The control method of the 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 dq coordinate system, note the predicted values of id and iq under zero voltage vector be point A (xi, yi), and the predicted values of id and iq under effective voltage vector be point B (x2, n): the reference current is located at point C (xo, yo\, the distance from point C to the line AB represents the minimum value of thecost 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:.vx,——=----J- (10)^2-^1 A-Athe equation of the line perpendicular to AB passing through point C is as follows:x2 x1-x2y=-----*+a---¾ (11)y2-A yi “Athe coordinates of the intersection point of the two lines can be obtained by solving equations (10) and (11) simultaneously:■pre _ *(A -¾)2 + ACA -½)2 + (xt -x2)(y1 -y0)(y2 -¾)d~duty (^-¾)2+(^-½)22 2 <12)■ pre ^0^-½) + 14(^-½) +(A-*o)(A-*2)(y2-Ji) / dm (^1-¾)2+04-½)2wherein, and represent the predicted values of the Jg-axis currents under modified duty cycle;withxo =i*d >y0 =CA =’ / +1 +TS + (oeLqiq+i]l Ldv _r_n j £+1_z a i j _ / )1 / / 1 / T. / 1 lq s L ^slq ^e^dd ci f J' ^qA = / / +1 +TS -l^+1 + a>eLqiqM\ILd= > / ' +TS-R^' -(DeWfyLqthe optimal duty cycle for the voltage vector can be determined based on the position of the intersection point.x — ipre'u adutyX) ^x2=1^(13)8. The control method of the 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, Gi, G3, and G4, using V4, V10, Vie, and V22 as boundaries; the virtual voltage vectors contained in each group areshown in the following table:Tab.2 Virtual Voltage Vector Partitioning RulesNum Voltage vectorsGi V22, V23. V24, V1, v2, V3, V4 G2 V4, Vs, Vs, v7, V8, ' Vs, V10 g3 V10, Vil, V12, V13, V14, V15 . Vis G4 V16, V17, Vis, V19, V20, V21 , V22by substituting Vi, V7, V13, and V19 into equation (5), the values of the cost function for each voltage vector are obtained: J(Vi), J(V?), / (Vb), / (Vw); the voltage vector Vut 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 Nind with the minimum cost function is selected, determining the second optimal group; assuming that N^nd 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 Vb, Nlnd Group V1M Nlnd Gi(Vi) V1 V2,V24 G3(Vi3) V11 V10, V12 V3 V2, V4 V13 V12, V14 V23 V.-2. V24 Vis V14, V16 G2(V7) Vs V4, Vs G4(Vw) V17 Vis, Vis V7 Vs, Vs V19 Vig, V20 Vs Vs, V10 V21 V20. V?2after 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. The control method of the 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; throughthe 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.11 11 24

Images