Five-phase motor fast pulse width modulation method based on multi-vector model predictive current control

Abstract

The invention provides a five-phase motor fast pulse width modulation method based on multi-vector model predictive current control, which is used for solving the problem that in the existing five-phase permanent magnet synchronous motor model predictive current control strategy, high control precision and low algorithm complexity cannot be achieved at the same time. The method comprises the following steps: establishing a motor model and a model prediction value evaluation function; in combination with the current error and coordinate transformation theory, updating the value evaluation function expression; the value evaluation function performs partial differential on the duty ratio to obtain a minimum value, and a virtual duty ratio is obtained; and the virtual duty ratio is reconstructed and then acts on an inverter to drive the five-phase motor to operate. The model predictive current control theory is adopted, the fast pulse width modulation strategy is combined, the optimal control duty ratio can be actively calculated in advance, and therefore faster dynamic response and lower overshoot are obtained compared with a traditional model predictive control strategy.

Description

Technical Field

[0001] This invention relates to the field of high-performance control of permanent magnet synchronous motors, specifically a fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control. Background Art

[0002] Permanent magnet synchronous motors (PMSMs) have become core power components in high-end industrial servo systems and new energy vehicles due to their high power density and efficiency. Continuous improvement in their control performance is key to driving development in these fields. To achieve high-performance control of PMSMs, in addition to traditional control methods (such as PI control), model predictive control (MMC) has become a research hotspot in academia and industry due to its advantages such as fast dynamic response, intuitive concepts, and ease of handling multiple variables. It is widely considered one of the mainstream technologies for next-generation motor control.

[0003] Currently, various modulation methods based on model predictive current control can be broadly categorized into single-vector and multi-vector methods. Single-vector modulation methods have a lower computational burden, but because only one basic voltage vector can be selected per cycle, their output is inherently discontinuous. This leads to errors between the selected optimal vector and the ideal target vector, resulting in significant current distortion and severely limited control accuracy. Multi-vector modulation methods analyze voltage vector space through geometric relationships, thus achieving higher control accuracy. However, their geometric solution process typically involves complex trigonometric function calculations or piecewise function judgments, resulting in high computational complexity and posing a significant challenge to the real-time computing capabilities of the controller. Summary of the Invention

[0004] The present invention aims to solve the major technical contradiction in the existing model predictive current control strategy for five-phase permanent magnet synchronous motors, which is that high control accuracy and low algorithm complexity cannot be achieved at the same time.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control, comprising:

[0006] S1: Collect the phase current and angle signals of the motor at the current moment, transform the phase current from the five-phase stationary coordinate system to the two-phase rotating coordinate system, and obtain the dq axis current signal;

[0007] S2: Based on the current time shaft and The duty cycle of the axis determines the current time. shaft and The voltage signal of the dq axis is obtained by combining it with the angle signal at the current moment;

[0008] S3: Based on the dq-axis current signal and the dq-axis voltage signal, the predicted value of the dq-axis current at the next moment is obtained by using the first-order Euler discretization equation of the permanent magnet synchronous motor.

[0009] S4: Determine the actual mechanical angular velocity based on the collected angle signal of the motor at the current moment;

[0010] S5: Determine the target value of the dq axis current;

[0011] S6: Calculate the current error on the dq axis at the next moment when the zero vector in the basic voltage vector is used as the input of the inverter switch, based on the predicted value of the dq axis current and the target value of the dq axis current.

[0012] S7: Construct a value assessment function based on the current dq-axis voltage signal and the current error caused on the dq-axis at the next moment;

[0013] S8: The value assessment function is extended based on the principle of one-time delay compensation for discrete systems, as well as the PWM modulation formula and the Parker transform formula;

[0014] S9: Apply the expanded value assessment function to the theoretically optimal value assessment function at the current moment. shaft duty cycle and Differentiate the duty cycles of each axis and set them to zero to obtain the theoretical optimum for the next time step. shaft and Axis duty cycle;

[0015] S10: Theoretically optimal for the next time step shaft and The axis duty cycle is converted to obtain a virtual duty cycle, and then the duty cycle is reconstructed to obtain the final duty cycle;

[0016] S11: The modulation pulse width is generated based on the final duty cycle, and the five-phase motor is driven to run through the inverter.

[0017] Compared with existing technologies, the significant advantages of this invention are as follows: This invention employs model predictive current control theory, enabling proactive and advance calculation of the optimal control duty cycle, thereby achieving a faster dynamic response and lower overshoot than traditional PI controllers. More importantly, thanks to the lightweight computation and nonlinear fitting characteristics of the ReLU function, this invention combines the lower algorithmic complexity of traditional single-vector model predictive current control with the higher control accuracy of traditional multi-vector model predictive current control, demonstrating promising application prospects in the field of low-cost, high-performance motor control. Attached Figure Description

[0018] Figure 1 This is a circuit topology diagram of the object of implementation of the present invention.

[0019] Figure 2 This is a flowchart of the fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to the present invention.

[0020] Figure 3 This is a comparison chart of experimental results between the method proposed in this invention and three other methods in the industry. The comparison dimensions are current circle, speed and torque. Figure 3 (a) and Figure 3 (b) is the traditional single-vector model predictive current control method. Figure 3 (c) is the traditional multi-vector model predictive current control method. Figure 3 (d) is the method proposed in this invention.

[0021] Figure 4 This is a comparison chart of experimental results between the method proposed in this invention and three other methods in the industry. The comparison dimension is the clock cycle required for MCU processing. Figure 4 (a) and Figure 4 (b) is the traditional single-vector model predictive current control method. Figure 4 (c) is the traditional multi-vector model predictive current control method. Figure 4 (d) is the method proposed in this invention. Detailed Implementation

[0022] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] This invention provides a fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control. Figure 1 The circuit topology diagram used in the verification of this invention is shown below. A five-phase inverter with a two-level voltage source drives a five-phase permanent magnet synchronous motor. The parameters of the five-phase motor are shown in Table 1.

[0024] Table 1

[0025]

[0026] Figure 2 This is a flowchart of the fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to the present invention. The specific process includes the following steps:

[0027] S1: Collect the current phase current of the motor at the current moment. and angle signal The phase current is represented by the following coordinate transformation theoretical formula. The dq-axis current signal is obtained by transforming from a five-phase stationary coordinate system to a two-phase rotating coordinate system. .

[0028]

[0029]

[0030] in , , , , , These represent the currents of phases A, B, C, D, and E at the current moment. , The fundamental space at the current time is respectively shaft current and shaft current, , The third harmonic space at the current time shaft current and shaft current, This is the zero-sequence current component. , These are the d-axis current and q-axis current at the current moment, respectively.

[0031] S2: Set the current time shaft and Duty cycle of the shaft Multiply by gain Get the current time shaft and shaft voltage signal The specific formula is as follows:

[0032]

[0033] Combined with the current angle signal The d-axis and q-axis voltage signals at the current moment are calculated. The specific formula is as follows:

[0034]

[0035] in, The magnitude of the basic voltage vector. , The result of finding the extremum of the partial derivative of the value assessment function at the current moment changes with each iteration. Initially, take... , In this embodiment, To account for the vector magnitude in the virtual voltage vector after the third spatial harmonic elimination method, its value is 0.5528. ,in It is 110V.

[0036] S3: Based on the fundamental theory of model prediction, the result obtained in S1... , And what S2 is looking for , Substituting these values ​​into the first-order Euler discretization equations of the permanent magnet synchronous motor, we obtain the predicted d-axis and q-axis current values ​​for the next time step. , The specific formula is as follows:

[0037] , In the formula, These are the predicted values ​​of the dq-axis current at the next time step. These are the dq-axis currents at the current time. These are the dq-axis voltages at the current time. , These represent the d-axis and q-axis inductances, respectively. Let be the electric angular velocity at the current moment. Indicates phase resistance. Indicates the sampling period. This indicates the magnetic flux linkage of a permanent magnet.

[0038] S4: The motor angle signal acquired in step S1 The actual mechanical angular velocity is obtained by performing the following calculations. .

[0039]

[0040]

[0041] in Let be the electric angular velocity at the current moment. Let be the mechanical angular velocity at the current moment. The number of pole pairs of the motor is given in this embodiment. .

[0042] S5: For the q-axis: Set the target machine speed With actual mechanical speed The difference is calculated and input to the PI controller, which then outputs the target current value. For the d-axis: use Strategy.

[0043] S6: Obtain the predicted d-axis and q-axis current values ​​from S3. , And the target values ​​of the d-axis and q-axis currents obtained from S5. and Perform the following calculations:

[0044]

[0045] in, , These represent the d-axis and q-axis inductance, respectively. Indicates phase resistance. Indicates the sampling period. This indicates the magnetic flux linkage of a permanent magnet. and These represent the current errors that will occur on the d-axis and q-axis at the next time step when the zero vector in the basic voltage vector is used as input. In this embodiment, the zero vector is the zero vector in the virtual voltage vector after considering the third spatial harmonic elimination method. mH, mH, , , .

[0046] S7: Construct a value assessment function based on the current dq-axis voltage signal and the current error caused on the dq-axis at the next moment:

[0047]

[0048] in This is the symbol for the value assessment function in the predictive current control of this scheme model.

[0049] S8: The value evaluation function is extended based on the principle of one-beat delay compensation for discrete systems, as well as the PWM modulation formula and the Parker transform formula:

[0050]

[0051]

[0052]

[0053] S9: The value assessment function in S8 against and By taking partial differentials separately and setting them to zero, we obtain the optimal solution. and The expression is as follows:

[0054]

[0055]

[0056] in For the electric angle at the next moment, and They represent the theoretical optimal values ​​at the next time step, respectively. shaft and Axis duty cycle.

[0057] S10: Convert the duty cycle expression obtained in S9 and Perform duty cycle conversion to obtain virtual duty cycle :

[0059]

[0060] The duty cycle is then reconstructed to obtain the final duty cycle. .

[0061]

[0062]

[0063] in This indicates a virtual duty cycle that has only undergone duty cycle conversion but not duty cycle reconstruction. This indicates the duty cycle that can be used for the final pulse width modulation after reconstruction. This indicates the zero vector duty cycle. The function is a maximum function. The function is a local minimum function. The function is a rectified function that truncates all negative input values ​​to zero, while leaving positive input values ​​unchanged. Its core idea is to introduce nonlinearity while maintaining computational simplicity. The specific expression of the function is as follows:

[0064]

[0065] S11: Based on the final duty cycle The modulation pulse width is generated, and then the five-phase motor is driven by the inverter.

[0066] This invention reduces the current distortion from 4.11% to 1.48% and the algorithm time cost from 17.23 microseconds to 3.01 microseconds by analyzing mathematical theory to solve the accurate duty cycle predicted by the multi-vector model. It has good application prospects in the field of low-cost, high-performance five-phase motor control.

[0067] Example

[0068] Under completely identical motor operating conditions Figure 3 The experimental results of four different model-based predictive current control methods are compared, with the comparison dimensions being, in order, the fundamental and harmonic current circles, rotational speed, and torque. Figure 3 (a) and Figure 3 (b) is the traditional single-vector model predictive current control modulation method. Figure 3 (c) is the traditional multi-vector model predictive current control modulation method. Figure 3 (d) shows the fast pulse width modulation method proposed in this invention. The figure summarizes that the performance of methods (a), (b), (c), and (d) improves to varying degrees in that order. Methods (a) and (b) perform the worst because the single-vector modulation strategy can only select one vector within a control cycle, making it impossible to synthesize vectors in arbitrary directions, resulting in higher current ripple and torque pulsation. Method (c) is a traditional multi-vector modulation strategy based on geometric approximation, which is slightly better than methods (a) and (b) because it uses multiple basic voltage vectors in the voltage vector space for vector synthesis, thus providing higher degrees of freedom to approximate the target voltage vector. Method (d) performs the best because the fast pulse width modulation method for five-phase motors based on multi-vector model predictive current control proposed in this invention is essentially an accurate solution to the duty cycle expression through analytical methods, which has inherent advantages over the geometric approximation method, thus exhibiting optimal output characteristics under the same operating conditions.

[0069] Figure 4 The experimental results comparing the time costs of the four methods in actual deployment are shown in the chart. The hardware solution uses a TMS320F28379D MCU with a clock frequency of 200MHz. Methods (a) and (b) have lower computational requirements due to their single-vector computational characteristics. Method (c) uses a multi-vector modulation strategy based on traditional geometric approximation, and its deployment process involves complex trigonometric function calculations and multiple loop judgment statements, thus requiring extremely high computational power. Method (d), benefiting from its simple duty cycle analytical expression, still has the lowest computational power requirement while having significantly better output characteristics than methods (a), (b), and (c). In particular, among the multi-vector model predictive current control modulation methods, the code execution time cost of method (d) is only 17.47% of that of method (c).

[0070] Therefore, the fast pulse width modulation method for five-phase motors based on multi-vector model predictive current control proposed in this invention can greatly reduce the algorithm complexity while improving the existing motor control effect, and has good prospects for industrial application, making it suitable for widespread promotion.

Claims

1. A fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control, characterized in that, include: S1: Collect the phase current and angle signals of the motor at the current moment, transform the phase current from the five-phase stationary coordinate system to the two-phase rotating coordinate system, and obtain the dq axis current signal; S2: Based on the current time shaft and The duty cycle of the axis determines the current time. shaft and The voltage signal of the dq axis is obtained by combining it with the angle signal at the current moment; S3: Based on the dq-axis current signal and the dq-axis voltage signal, the predicted value of the dq-axis current at the next moment is obtained by using the first-order Euler discretization equation of the permanent magnet synchronous motor. S4: Determine the actual mechanical angular velocity based on the collected angle signal of the motor at the current moment; S5: Determine the target value of the dq axis current; S6: Calculate the current error on the dq axis at the next moment when the zero vector in the basic voltage vector is used as the input of the inverter switch, based on the predicted value of the dq axis current and the target value of the dq axis current; S7: Construct a value assessment function based on the current dq-axis voltage signal and the current error caused on the dq-axis at the next moment; S8: The value assessment function is extended based on the principle of one-time delay compensation for discrete systems, as well as the PWM modulation formula and the Parker transform formula; S9: Apply the expanded value assessment function to the current moment. shaft duty cycle and Differentiate the duty cycles of each axis and set them to zero to obtain the theoretical optimum for the next time step. shaft and Axis duty cycle; S10: Theoretically optimal for the next time step shaft and The axis duty cycle is converted to obtain a virtual duty cycle, and then the duty cycle is reconstructed to obtain the final duty cycle. S11: The modulation pulse width is generated based on the final duty cycle, and the five-phase motor is driven to run through the inverter.

2. The fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to claim 1, characterized in that, The specific formula for transforming the phase current from the five-phase stationary coordinate system to the two-phase rotating coordinate system to obtain the dq-axis current signal is as follows: ， ， in , , , , , These represent the currents of phases A, B, C, D, and E at the current moment. , The fundamental space at the current time is respectively shaft current and shaft current, , The third harmonic space at the current time shaft current and shaft current, This is the zero-sequence current component. , These are the d-axis current and q-axis current at the current moment, respectively. angle signal.

3. The fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to claim 1, characterized in that, According to the current time shaft and The duty cycle of the axis determines the current time. shaft and The specific formula for the voltage signal of the shaft is: ， The specific formula for obtaining the dq-axis voltage signal at the current moment, combining the angle signal at the current moment, is as follows: ， in, The magnitude of the basic voltage vector. , , Theoretically optimal at the current time shaft duty cycle and Axis duty cycle, , These are the d-axis and q-axis voltage signals at the current time. The current electrical angle signal is given at the initial moment. , .

4. The fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to claim 1, characterized in that, Based on the dq-axis current and voltage signals, the specific formula for obtaining the predicted value of the dq-axis current at the next moment using the first-order Euler discretization equation of the permanent magnet synchronous motor is as follows: ， In the formula, These are the predicted values ​​of the dq-axis current at the next time step. These are the dq-axis currents at the current time. These are the dq-axis voltages at the current time. , These represent the d-axis and q-axis inductances, respectively. Let be the electric angular velocity at the current moment. Indicates phase resistance. Indicates the sampling period. This indicates the magnetic flux linkage of a permanent magnet.

5. The fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to claim 1, characterized in that, The specific formula for determining the actual mechanical angular velocity based on the collected angle signal of the motor at the current moment is as follows: ， ， in, Let be the electric angular velocity at the current moment. Let be the mechanical angular velocity at the current moment. The electric angle at the current moment, This represents the number of pole pairs of the motor.

6. The fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to claim 1, characterized in that, The method for determining the target value of the q-axis current is as follows: target mechanical speed With actual mechanical speed The difference is calculated and input to the PI controller, which then outputs the target q-axis current value. ; d-axis current target value .

7. The fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to claim 1, characterized in that, Based on the predicted and target values ​​of the dq-axis current, the specific formula for calculating the current error on the dq-axis at the next moment when the zero vector in the basic voltage vector is used as the input to the inverter switch is as follows: ， in, , These represent the d-axis and q-axis inductance, respectively. Indicates phase resistance. Indicates the sampling period. Indicates permanent magnet flux linkage. and These represent the current errors on the d-axis and q-axis at the next moment when the zero vector from the basic voltage vector is used as the input to the inverter switch. The target value for the q-axis current. The target value for the d-axis current. Let be the electric angular velocity at the current moment. These are the predicted current values ​​for the d-axis and q-axis at the next time step, respectively.

8. The fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to claim 7, characterized in that, The value assessment function, constructed based on the current dq-axis voltage signal and the current error caused on the dq-axis at the next moment, is as follows: ， In the formula, , These are the d-axis and q-axis voltage signals at the current moment, respectively.

9. The fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to claim 8, characterized in that, Based on the principle of one-time delay compensation for discrete systems, as well as the PWM modulation formula and the Park transform formula, the value evaluation function is extended. The extended value evaluation function is as follows: ； In the formula, and They represent the theoretical optimal values ​​at the next time step, respectively. shaft and Axis duty cycle, For the electrical angle signal at the next moment, This represents the magnitude of the basic voltage vector.

10. The fast pulse width modulation method for a five-phase motor based on multi-vector model predictive current control according to claim 9, characterized in that, The extended value assessment function is theoretically optimal for the current moment. shaft duty cycle and Differentiate the duty cycles of each axis and set them to zero to obtain the theoretical optimum for the next time step. shaft and The specific formula for the shaft duty cycle is: ， 。

Images