A method for predicting deadbeat current control using a three-vector model of PMSM based on geometric analysis

Abstract

This invention provides a three-vector model predictive deadbeat current control method for PMSM based on geometric analysis. The method mainly includes the following steps: Step 1: Establish a mathematical model of the PMSM motor, use this model to obtain the current prediction quantity for model predictive current control, and then use deadbeat control to convert the current prediction quantity into a voltage prediction quantity; Step 2: Set the cost function in model predictive control to the root mean square form, and establish a relationship between the cost function value and the vectors in the voltage vector diagram; Step 3: Calculate the duty cycle of the three vectors' action time for each cycle by analyzing the geometric relationship of the voltage vector diagram; Step 4: Pre-modulate or compensate the voltage of the over-modulated part to ensure stable motor operation and reduce the error during over-modulation.

Description

Technical Field

[0001] This invention belongs to the field of PMSM control, specifically relating to a method for predicting deadbeat current control of PMSM using a three-vector model based on geometric analysis. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) are characterized by their simple structure, high efficiency, high power density, and reliable operation, and are widely used in various fields such as industrial and agricultural production and aerospace. Model predictive control (MPC), due to its ease of incorporating multivariable constraints and excellent control performance, is now widely used in power converters, wind energy conversion systems, and motors. Based on different optimization methods and operating modes, MPC is divided into Continuous Control Set MPC (CCS-MPC) and Finite Control Set MPC (FCS-MPC). FCS-MPC, which considers both objective optimization and switching state decision-making processes holistically, has advantages such as wide applicability and ease of application.

[0003] FCS-MPC calculates the cost function of seven basic voltage vectors within one cycle, selects the voltage vector with the smallest cost function value for output, and continuously optimizes the motor's operating state to approach the set target state. However, single-voltage vector model predictive control can lead to significant torque ripple and current harmonics. To address this issue, this invention proposes a deadbeat-free current control method for PMSM based on geometric analysis using a three-vector model predictive control. Summary of the Invention

[0004] To address the significant torque ripple and current harmonics inherent in single-vector model predictive current control, this invention provides a deadbeat-free PMSM three-vector model predictive current control method based on geometric analysis. The proposed method does not increase computational complexity excessively and effectively reduces the large torque ripple and current harmonics associated with single-vector model predictive control.

[0005] The technical solution of the present invention is as follows:

[0006] This invention proposes a deadbeat-free current control method based on a geometric analysis-based three-vector model of a PMSM (Predictive PMSM), which includes the following steps:

[0007] 1) Establish a mathematical model of the PMSM motor, use the model to obtain the current prediction quantity for model-predicted current control, and then use deadbeat control to convert the current prediction quantity into the voltage prediction quantity.

[0008] 2) Set the cost function in model predictive control to the root mean square form, and establish a relationship between the cost function value and the vector in the voltage vector diagram;

[0009] 3) Determine the two non-zero vectors of the synthesized target voltage vector, and calculate the duty cycle of the two non-zero vectors and one zero vector in each cycle by analyzing the geometric relationship of the voltage vector diagram;

[0010] 4) Pre-modulate or compensate the voltage of the over-modulated part to ensure stable operation of the motor and reduce the error during over-modulation.

[0011] As a preferred embodiment of the present invention, the establishment of the mathematical model of the PMSM motor in step 1) specifically refers to: the direct-axis inductance (L) of the surface-mounted permanent magnet synchronous motor. d ) equals quadrature axis inductance (L q Therefore, let L d =L q =L, where L represents the quadrature-direct axis inductance;

[0012] The mathematical model of the motor is established in the dq coordinate system as follows:

[0013]

[0014] In the formula, i d i q These are the d-axis and q-axis components of the stator current, respectively; u d u q ω represents the d-axis and q-axis components of the stator voltage, respectively; R is the stator resistance; L is the stator inductance; ω e Ψ is the rotor's electrical angular velocity. f It is a permanent magnet flux linkage.

[0015] Furthermore, the current prediction quantity for model-predicted current control obtained using this model in step 1) specifically includes:

[0016] Given a sufficiently small sampling period Ts, we can discretize equation (1) using Euler's formula to obtain:

[0017]

[0018] In the formula, i d (k), i q (k) represents the sampled values ​​of the d-axis and q-axis components of the stator current at the current moment; u d (k), u q (k) represents the sampled values ​​of the d-axis and q-axis components of the stator voltage at the current moment, where the stator voltage is determined by the switching state of the inverter. Different inverter switching combinations will result in different u values. d(k), u q (k); ω e (k) represents the sampled value of the rotor's electric angular velocity at the current moment; T S Sampling time; These are the predicted values ​​of the stator current d-axis and q-axis components at the next moment.

[0019] Preferably, in step 1), the method of converting the current prediction into a voltage prediction using deadbeat control is as follows: Based on the principle of deadbeat control, Ts is sufficiently small, so the current prediction value at time k+1 is equal to the current reference value at time k, i.e.:

[0020]

[0021] In the formula, i dref (k), i qref (k) are the reference values ​​of the stator current d and q axis components at time k, respectively; according to equations (2) and (3), the ideal stator voltage vector is obtained as:

[0022]

[0023] Therefore, the predicted current is transformed into the predicted voltage, and the dq-axis components of the ideal voltage vector are obtained.

[0024] According to a preferred embodiment of the present invention, step 2) specifically comprises:

[0025] First, the cost function in model predictive control is set to the root mean square form, i.e., cost function j:

[0026]

[0027] In the formula, u dn The stator voltage vector provided for different switching combinations of the inverter, n takes the value 0, 1, ..., 7; the stator voltage vector with the minimum cost function value j is the voltage vector to be selected. This voltage vector corresponds to different switching states of the inverter, thereby generating pulse signals to guide the switching transistors to operate.

[0028] Secondly, the cost function value is linked to the vectors in the voltage vector diagram. The cost function value of the non-zero basic voltage vector in the voltage vector diagram represents the degree of proximity to the target voltage vector and also represents the vector length corresponding to the difference between the two vectors.

[0029] According to a preferred embodiment of the present invention, the determination of the two non-zero vectors of the synthesized target voltage vector in step 3) specifically refers to:

[0030] The cost function value represents the vector length corresponding to the difference between two vectors. Therefore, the vector length corresponding to the difference between two vectors can be used to represent the magnitude of the cost function. Thus, the cost function value can be obtained using the cosine theorem.

[0031]

[0032] Where j1 is the cost function value of the basic voltage vector u1; similarly, the cost function values ​​of the other 5 non-zero vectors are obtained. Obviously, u1 = u2 = ... u6. Only the angle ρ between the non-zero basic voltage vector and the target voltage vector affects the value of the cost function j. In the range [0, 180], the cosine function is monotonically decreasing. Obviously, the angle between the two non-zero basic voltage vectors in the sector where the target voltage vector is located and the target voltage vector is the smallest, and its cost function is also the smallest. Therefore, when performing the cost function cyclic judgment in each cycle, the cost function is calculated from the non-zero vector u1. n For n = 1, 2, ..., 6, select the two basic voltage vectors with the smallest cost function values, which are also the two non-zero voltage vectors in the sector where the target voltage vector is located.

[0033] Furthermore, in step 3), the duty cycle of the two non-zero vectors and one zero vector in each cycle is calculated by analyzing the geometric relationship of the voltage vector diagram. Specifically:

[0034] ① Calculation of zero vector duty cycle:

[0035] The zero vector duty cycle can be obtained by combining the area of ​​the triangle formed by the vectors in the voltage vector diagram:

[0036]

[0037] j1 and j2 are the minimum and second minimum values ​​of the cost function, respectively; U dc is the DC power supply voltage of the inverter; p is half the perimeter of the vector composite triangle.

[0038] ② Calculation of duty cycle of non-zero vectors:

[0039] The duty cycles of two non-zero vectors are obtained using the law of cosines:

[0040]

[0041]

[0042] According to a preferred embodiment of the present invention, step 4) specifically involves: judging the target voltage obtained by deadbeat control; if the target voltage u *If the voltage exceeds the modulation range, it is pre-modulated to the range of the vector synthesis triangle to ensure the normal operation of the motor. This also ensures that the adjusted voltage vector can be modulated correctly, and normal modulation guarantees stable motor operation. Furthermore, it only takes effect during over-modulation and will not affect the overall performance of the control system. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of a motor control system;

[0044] Figure 2 This is a schematic diagram of the process flow of the method of the present invention;

[0045] Figure 3 This is a schematic diagram of a voltage vector;

[0046] Figure 4 An auxiliary diagram used for calculating the duty cycle;

[0047] Figure 5 This is a schematic diagram of overmodulation;

[0048] Figure 6 This is a simulation experiment speed curve diagram of the present invention;

[0049] Figure 7 This is a schematic diagram showing the comparison of steady-state torque in the simulation experiment of this invention;

[0050] Figure 8 This is a schematic diagram comparing the steady-state current Fourier analysis of the present invention. Detailed Implementation

[0051] The present invention will be further described and illustrated below with reference to figures and specific embodiments. The embodiments described are merely examples of the content of this disclosure and do not limit the scope of the invention. The technical features of each embodiment in the present invention can be combined accordingly, provided that there is no mutual conflict.

[0052] according to Figure 1 The entire motor control system flow can be seen as follows:

[0053] ① Collect motor operating status information, including rotor rotational angular velocity ω, position θ, etc. The rotor position information is used as the basis for coordinate transformation. The rotor rotational angular velocity ω is converted into speed per minute n (i.e., speed information n) and sent to the speed loop.

[0054] ②Based on the current rotational speed information n and the expected reference rotational speed n ref The difference is fed into the PI controller so that the actual speed can track the reference speed, which is commonly referred to as the speed loop;

[0055] ③ The output of the speed loop PI regulator can be used as a reference value for the q-axis current in the next stage. qrefThe d-axis current uses i d =0 control mode, so i dref =0, change i qref i dref The dq components id and iq of the current are fed into the three-vector model prediction deadbeat current control module proposed in this invention, and then the inverter control pulse is output after calculation to control the motor speed.

[0056] The method for predicting the deadbeat current control module using a three-vector model is as follows: Figure 2 As shown:

[0057] ① Receive the current operating information of the motor and the reference current obtained by the speed loop PI controller, and then calculate the ideal target voltage through the deadbeat control principle;

[0058] ② The cost function between the non-zero basic voltage and the ideal target voltage vector is calculated in 6 cycles. Then, the duty cycle of the zero vector and the two non-zero vectors is calculated by the duty cycle. Finally, a pulse signal is generated to control the inverter switch.

[0059] The specific implementation steps of the algorithm are as follows:

[0060] 1) The mathematical model of the PMSM motor is established as follows. This embodiment uses a surface-mounted permanent magnet synchronous motor as the example, that is, the default direct-axis inductance (L) is assumed to be... d ) equals quadrature axis inductance (L q Therefore, let L d =L q =L, and L will be used to replace the direct and quadrature axis inductances thereafter.

[0061] The mathematical model of the motor is established in the dq coordinate system as follows:

[0062]

[0063] In the formula, i d i q These are the d-axis and q-axis components of the stator current, respectively; u d u q ω represents the d-axis and q-axis components of the stator voltage, respectively; R is the stator resistance; L is the stator inductance; ω e Ψ is the rotor's electrical angular velocity. f It is a permanent magnet flux linkage.

[0064] Secondly, a model predictive control system was established based on the mathematical model of the PMSM motor:

[0065] Assuming the sampling period Ts is sufficiently small, discretizing the above equation using Euler's formula yields:

[0066]

[0067] In the formula, i d (k), i q (k) represents the sampled values ​​of the d-axis and q-axis components of the stator current at the current moment; u d (k), u q (k) represents the d-axis and q-axis component samples of the stator voltage at the current moment (the stator voltage is determined by the switching state of the inverter; different inverter switching combinations will result in different u values). d (k), u q (k). );ω e (k) represents the sampled value of the rotor's electric angular velocity at the current moment; T S Sampling time; These are the predicted values ​​of the stator current d-axis and q-axis components at the next moment.

[0068] Based on the principle of deadbeat control, Ts is sufficiently small, so that the predicted current value at time k+1 is equal to the reference current value at time k, that is:

[0069]

[0070] In the formula, i dref (k), i qref (k) represents the reference values ​​of the stator current d-axis and q-axis components at time k, respectively; based on the above analysis, the ideal stator voltage vector can be obtained as:

[0071]

[0072] 2) Set the cost function in model predictive control to the root mean square form, i.e., cost function j:

[0073]

[0074] In the formula, u dn The stator voltage vector provided for different switching combinations of the inverter, where n takes values ​​of 0, 1, ..., 7. The stator voltage vector whose cost function value j is minimized is the selected voltage vector. This voltage vector corresponds to different switching states of the inverter, thereby generating pulse signals to guide the switching transistors to operate.

[0075] Secondly, a relationship is established between the cost function value and the vectors in the voltage vector diagram:

[0076] By analyzing the position of the target voltage vector, two non-zero voltage vectors are determined for synthesizing the target voltage vector.

[0077] Figure 3This is a schematic diagram of voltage vectors, where u* is the target voltage vector; u1 and u2 are two non-zero basic voltage vectors of the sector where the target voltage vector is located; ρ is the angle between the target voltage vector and the corresponding basic voltage vector; A, B, C, and O represent the endpoints of each vector.

[0078] Combining cost function and Figure 4 It can be confirmed that:

[0079]

[0080] Where j1 is the cost function value of the basic voltage vector u1; similarly, the length of vector BO, |BO|, can be obtained as j2; therefore, the cost function value of a non-zero basic voltage vector not only represents the degree of proximity to the target voltage vector, but also represents the vector length corresponding to the difference between the two vectors.

[0081] 3) First, determine the two non-zero vectors of the synthesized target voltage vector. From the above analysis, we know that the cost function value represents the vector length corresponding to the difference between the two vectors. Therefore, the vector length corresponding to the difference between the two vectors can also be used to represent the magnitude of the cost function. Thus, applying the law of cosines to ∠OAC in △OAC yields:

[0082]

[0083] Similarly, we can obtain the magnitudes of the cost function values ​​for the six non-zero vectors n = 1, ..., 6. Clearly, u1 = u2 = ..., u6. Only the angle ρ between the non-zero basic voltage vector and the target voltage vector affects the magnitude of the cost function j. Within the range [0, 180], the cosine function is monotonically decreasing. Obviously, the angle between the two non-zero basic voltage vectors in the sector containing the target voltage vector and the target voltage vector is the smallest, and its cost function is also the smallest. Therefore, during the cyclic judgment of the cost function in each cycle, the non-zero vector u... n For n = 1, 2, ..., 6, select the two basic voltage vectors with the smallest cost function values, which are also the two non-zero basic voltage vectors of the sector where the target voltage vector is located.

[0084] Secondly, the duty cycle of each cycle of the three vectors (two non-zero vectors and one zero vector) is calculated by analyzing the geometric relationship of the voltage vector diagram.

[0085] The duty cycles of non-zero and zero vectors are obtained by performing geometric analysis in △ABC.

[0086] Figure 4 An auxiliary diagram used to calculate the duty cycle, where AC, AB, and AS are respectively Figure 4Let u1, u2, u* be the points; connect BC; E and D are the intersections of the line passing through S and parallel to BC with AC and AB, respectively; G is the intersection of the line passing through S and parallel to AB with AC; F is the intersection of the line passing through S and parallel to AC with AB.

[0087] ① Calculation of zero vector duty cycle:

[0088] According to the principle of vector composition, |AD| / |AB| is the total duty cycle of non-zero vectors; |DB| / |AB| is the duty cycle of zero vectors.

[0089] Where |DB| / |AB| can be converted into the ratio of the height from S to BC to the height from A to BC, i.e., the zero vector duty cycle:

[0090]

[0091] Where the height h from S to BC S→BC It can be calculated using the formula for the area of ​​a triangle:

[0092]

[0093] From the analysis in the previous section, we know that |SC|=j1、|SB|=j2, where j1 and j2 are the minimum and second minimum values ​​of the cost function, respectively; |BC|=|AB|=|AC|=2U dc / 3, U dc The inverter's DC power supply voltage; combined with equation (8), the zero vector duty cycle can be calculated as:

[0094]

[0095] ② Calculation of duty cycle of non-zero vectors:

[0096] According to the principle of vector composition, D1 = |AG| / |AC| is the duty cycle of the basic vector u1; D2 = |AF| / |AB| is the duty cycle of the basic vector u2, and D1 + D2 = 1 - D0; Figure 5 In the triangle, △ABC, △SFD, and △SGE are equilateral triangles, and quadrilateral AGSF is a parallelogram. Therefore, |AG|=|SF|=|SD|、|AF|=|SE|=|SG|.

[0097] Apply the Law of Cosines to ∠SDB and ∠SEC in △SDB and △OEC respectively:

[0098]

[0099] Where |DB|=|EC|=(1-D0)*2U dc / 3;|SD| / |AC|=D1;|SE| / |AB|=D2;|SC|=j1、|SB|=j2;∠SDB=∠SEC=120°;Subtracting the two equations in equation (10) yields the duty cycles D1 and D2 of the basic vectors u1 and u2:

[0100]

[0101]

[0102] Thus, the duty cycles D0, D1, and D2 of the three vectors (two non-zero vectors and one zero vector) used to synthesize the target voltage vector can be obtained for each cycle.

[0103] 4) This invention employs deadbeat current control, directly using the stator voltage to calculate the cost function. However, this voltage can exceed the modulation range, therefore, it is necessary to judge the target voltage obtained from deadbeat control. For example, if the target voltage u... * If the voltage exceeds the modulation range, it needs to be pre-modulated to ensure the normal operation of the motor.

[0104] Figure 5 This is a schematic diagram of overmodulation, where AC, AB, and AO are respectively... Figure 3 The values ​​u1, u2, and u* are shown. At this point, AO is outside the overmodulation range, and D is the intersection of AO and BC. To ensure that the voltage is not overmodulated, vector AD is made the new target voltage vector. Therefore:

[0105]

[0106] The above describes the specific method of this invention. The following figures illustrate the effectiveness and superiority of the method proposed in this invention.

[0107] Figure 6 Figure (a) shows the acceleration and deceleration speed curves under a load of 3 N·m. The initial given speed is 1500 r / min. At 0.4 s, the given speed is changed to 750 r / min, and at 0.8 s, the given speed is changed to 1800 r / min. It can be seen from the figure that the motor response speed is very fast. After the given speed changes, it basically reaches equilibrium in about 0.02 s, which shows that the algorithm proposed in this invention enables the motor to maintain a good dynamic response. Figure (b) shows the acceleration and deceleration speed curves under a given speed of 1500 r / min. The initial load is set to 1 N·m. At 0.4 s, the load is set to 3 N·m, and at 0.8 s, the load is set to 0 N·m. It can be seen that the speed hardly changes, which shows the good dynamic adjustment capability of the method described in this invention.

[0108] Figure 7To simulate the steady-state torque diagram, the experimental conditions were set as a load of 3 N·m and a given speed of 1500 r / min. Figure (a) shows the steady-state torque diagram of the traditional model predictive control method, and Figure (b) shows the steady-state torque diagram of the method of this invention. It can be seen that the torque fluctuation range of the traditional method is between 3.25 and 5.00 N·m, while the torque fluctuation range of the method of this invention is between 3.7 and 4.7 N·m. To more accurately determine the magnitude of the torque ripple, the standard deviation of the collected data was calculated. It was found that the standard deviation of the traditional method was 0.3146 N·m, while the standard deviation of the method of this invention was 0.1954 N·m, indicating that the method of this invention has better steady-state torque performance.

[0109] Figure 8 To simulate the steady-state current and Fourier analysis plots, the experimental conditions were set as follows: load 3 N·m and given rotational speed 1500 r / min. Figure (a) shows the steady-state current plot of the traditional model predictive control method; Figure (b) shows the Fourier analysis plot of the steady-state current of the traditional model predictive control method; Figure (c) shows the steady-state current plot of the method of this invention; and Figure (d) shows the Fourier analysis plot of the steady-state current of the method of this invention. It can be seen that the harmonic content of the steady-state current of the traditional method is 10.46%, while the harmonic content of the steady-state current of the method of this invention is 6.68%, indicating that the method of this invention has better sinusoidal steady-state current performance.

[0110] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting deadbeat current control using a three-vector model of a PMSM based on geometric analysis, characterized in that, Includes the following steps: 1) Establish a mathematical model of the PMSM motor, use the model to obtain the current prediction quantity for model-predicted current control, and then use deadbeat control to convert the current prediction quantity into the voltage prediction quantity. 2) Set the cost function in model predictive control to the root mean square form, and establish a relationship between the cost function value and the vector in the voltage vector diagram; 3) Determine the two non-zero vectors of the synthesized target voltage vector, and calculate the duty cycle of the two non-zero vectors and one zero vector in each cycle by analyzing the geometric relationship of the voltage vector diagram; The two non-zero vectors for determining the synthesized target voltage vector in step 3) are specifically as follows: The cost function value represents the vector length corresponding to the difference between two vectors. Therefore, the vector length corresponding to the difference between two vectors can be used to represent the magnitude of the cost function. Thus, the cost function value can be obtained using the cosine theorem. ； Where j1 is the cost function value of the basic voltage vector u1; similarly, the cost function values ​​of the other 5 non-zero vectors are obtained. Obviously, u1=u2=…u6. Only the angle ρ between the non-zero basic voltage vector and the target voltage vector affects the value of the cost function j. In the range [0,180], the cosine function is monotonically decreasing. Obviously, the angle between the two non-zero basic voltage vectors in the sector where the target voltage vector is located and the target voltage vector is the smallest, and its cost function is also the smallest. Therefore, when performing the cost function cyclic judgment in each cycle, the cost function is calculated from the non-zero vector u1. n For n=1,2,…,6, select the two basic voltage vectors with the smallest cost function values, which are also the two non-zero voltage vectors in the sector where the target voltage vector is located; Step 3) describes calculating the duty cycle of the two non-zero vectors and one zero vector during each cycle by analyzing the geometric relationship of the voltage vector diagram. Specifically: ① Calculation of zero vector duty cycle: The zero vector duty cycle can be obtained by combining the area of ​​the triangle formed by the vectors in the voltage vector diagram: (9) j1 and j2 are the minimum and second minimum values ​​of the cost function, respectively; U dc This refers to the DC power supply voltage of the inverter. It is half the perimeter of the vector-composed triangle; ② Calculation of duty cycle of non-zero vectors: The duty cycles of two non-zero vectors are obtained using the law of cosines: (11) (12) 4) Pre-modulate or compensate the voltage of the over-modulated part to ensure stable operation of the motor and reduce the error during over-modulation.

2. The method according to claim 1, characterized in that, The establishment of the PMSM motor mathematical model described in step 1) is specifically as follows: For surface-mounted permanent magnet synchronous motors, the direct-axis inductance L d Equal to quadrature axis inductance L q Therefore, let L d =L q =L, where L represents the quadrature-direct axis inductance; The mathematical model of the motor is established in the dq coordinate system as follows: (1) In the formula, i d i q These are the d-axis and q-axis components of the stator current, respectively; u d u q ω represents the d-axis and q-axis components of the stator voltage, respectively; R is the stator resistance; L is the stator inductance; ω e Ψ is the rotor's electrical angular velocity. f It is a permanent magnet flux linkage.

3. The method according to claim 2, characterized in that, Step 1) describes obtaining the predicted current for model-predicted current control using this model, specifically as follows: Given a sufficiently small sampling period Ts, we can discretize equation (1) using Euler's formula to obtain: (2) In the formula, i d (k), i q (k) represents the sampled values ​​of the d-axis and q-axis components of the stator current at the current moment; u d (k), u q (k) represents the sampled values ​​of the d-axis and q-axis components of the stator voltage at the current moment, where the stator voltage is determined by the switching state of the inverter. Different inverter switching combinations will result in different u values. d (k), u q (k); ω e (k) represents the sampled value of the rotor's electric angular velocity at the current moment; T S Sampling time; (k+1) (k+1) represents the predicted values ​​of the d-axis and q-axis components of the stator current at the next moment.

4. The method according to claim 3, characterized in that, Step 1) describes the reuse of deadbeat control to convert the current prediction into a voltage prediction, specifically as follows: Based on the principle of deadbeat control, Ts is sufficiently small, so that the predicted current value at time k+1 is equal to the reference current value at time k, that is: (3) In the formula, i dref (k), i qref (k) are the reference values ​​of the stator current d and q axis components at time k, respectively; according to equations (2) and (3), the ideal stator voltage vector is obtained as: (4) Therefore, the predicted current is transformed into the predicted voltage, and the dq-axis components of the ideal voltage vector are obtained.

5. The method according to claim 1, characterized in that, Step 2) specifically refers to: First, the cost function in model predictive control is set to the root mean square form, i.e., cost function j: (5) In the formula, u dn The stator voltage vector provided for different switching combinations of the inverter, where n takes values ​​of 0, 1, ..., 7; The stator voltage vector that minimizes the cost function value j is the voltage vector to be selected. This voltage vector corresponds to different switching states of the inverter, thereby generating pulse signals to guide the switching transistors to operate. Secondly, the cost function value is linked to the vectors in the voltage vector diagram. The cost function value of the non-zero basic voltage vector in the voltage vector diagram represents the degree of proximity to the target voltage vector and also represents the vector length corresponding to the difference between the two vectors.

6. The method according to claim 1, characterized in that, Step 4) specifically involves: judging the target voltage obtained from the deadbeat control; if the target voltage u * If the voltage exceeds the modulation range, it will be pre-modulated to the range of the vector synthesis triangle to ensure the normal operation of the motor.

Images