Multi-vector model predictive control method based on incremental global optimization
Through the multi-vector model prediction control method with incremental global optimization, the problem of large calculation amount of multi-vector model prediction control and mismatch of magnetic flux parameters is solved, and higher prediction accuracy and robustness are achieved, and it is suitable for motor control of N*3 phase structure.
Patent Information
- Application Number
- CN202510747276.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-05
- Publication Date
- 2025-08-29
AI Technical Summary
The existing multi-vector model prediction control methods have large calculations, high model accuracy requirements, poor control performance due to mismatch of magnetic flux parameters, and the optimal synthesis scheme is not necessarily globally optimal.
The incremental global optimization method is adopted to discrete the current state equation through the Euler method, and a current prediction model is constructed, a virtual vector value function with the harmonic plane component of zero is set, an incremental voltage prediction model is derived, and a similar vector is selected for duty cycle allocation is selected to reduce the calculation amount and improve the prediction accuracy.
It reduces the calculation burden, improves the system robustness and prediction accuracy, reduces the impact of magnetic flux parameter mismatch, and expands the application range to N*3-phase structure motor control.
Smart Images
Figure CN120566982A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of power electronics and motor control, and in particular to a multi-vector model predictive control method based on incremental global optimization. Background Art
[0002] The torque ripple of the FCMPC method is greater than that of the FCMPC method, but the current quality is much better, especially in the no-load steady state, where the current ripple is very small. Therefore, FCS-MPCC also has good development prospects in industrial applications.
[0003] Finite control set model predictive current control can directly output the on-off signal of the control switch tube and has good dynamic performance, which can meet the requirements of permanent magnet synchronous motors for fast response. Multi-vector model predictive control expands the control set from point to surface, improving prediction accuracy and is a current research hotspot. At present, multi-vector control methods usually use a value function to select two adjacent optimal vectors and then allocate the duty cycle, resulting in the final composite solution not necessarily being the global optimal. Existing methods also use a value function to optimize after traversing the duty cycle allocation scheme, but this increases the computational burden of the processor. At the same time, because model prediction depends on model accuracy, mismatch of flux parameters will lead to prediction errors, greatly reducing control performance.
[0004] The improvement of the prediction accuracy of the existing multi-vector model predictive current control is accompanied by an increase in the amount of calculation, which places high demands on the accuracy of the model. Improving the system prediction accuracy with a lower computational burden and improving the system robustness are issues that need to be urgently addressed. Summary of the Invention
[0005] The purpose of the present invention is to overcome the defects existing in the above-mentioned background technology. The present invention proposes a multi-vector model predictive control method based on incremental global optimization to realize the control of permanent magnet synchronous motor. The reference voltage vector is obtained according to the incremental voltage prediction equation, thereby reducing the vector control set to reduce the amount of calculation; the calculation of the value function takes into account the result of the synthetic vector action after the duty cycle distribution, so that the obtained scheme is a global optimization result, thereby improving the prediction accuracy of the system.
[0006] The present invention adopts the following technical solutions to solve the above technical problems:
[0007] A multi-vector model predictive control method based on incremental global optimization is characterized in that the method controls a permanent magnet synchronous motor through the following steps:
[0008] Step S1, constructing a discretization and current prediction model: using the Euler method to discretize the dq axis current state equations to obtain a current prediction model including dq axis voltage components, quadrature and direct axis inductances, stator resistance, rotor electrical angular velocity, and permanent magnet flux parameters;
[0009] Step S2, setting the value function: using a virtual vector with zero harmonic plane component as the vector control set, and setting the value function to the sum of the squares of the differences between the dq-axis predicted current and the reference current;
[0010] Step S3, deriving an incremental equation and constructing a voltage prediction model: deriving a current prediction equation for the kth cycle, subtracting the prediction current equations of two adjacent cycles to obtain an incremental current prediction equation, setting the predicted current equal to the reference current, and obtaining an incremental voltage prediction model that eliminates the flux linkage term;
[0011] Step S4, vector selection: Calculate the vector angle of the reference voltage, select a vector close to the vector angle according to the vector set as the first vector of the synthetic reference vector, and narrow the selection range of the second vector to two adjacent vectors of the first vector;
[0012] Step S5, duty cycle allocation and synthesis: performing duty cycle allocation on the selected voltage vector scheme according to the Lagrange multiplier method to obtain the synthesized voltage vector and current prediction value after duty cycle allocation;
[0013] Step S6, updating the value function and determining the solution: updating the value function to the difference between the predicted current and the reference current under the action of the synthetic vector, and obtaining the global optimized vector solution after the duty cycle is allocated.
[0014] Furthermore, the step S1 is specifically as follows:
[0015] The current prediction model is obtained by discretizing the dq axis current state equation using the Euler method:
[0016]
[0017] Among them, T s is the control period, u d k 、u q k are the voltage components of the kth cycle of the dq axis, i d k 、i q k are the k-th cycle sampling currents of dq axis, i d k+1 、i q k+1 They are the predicted current of dq axis k+1 period, L q , L d is the DC-axis inductance, R s is the stator resistance, ω e is the rotor electrical angular velocity, is the permanent magnet flux.
[0018] Furthermore, the step S2 is specifically as follows:
[0019] A virtual vector with zero harmonic plane components is used as the vector control set, and the cost function is the sum of the squares of the differences between the dq-axis predicted current and the reference current:
[0020] J=(i d * -i d k+1 ) 2 +(i q * -i q k+1 ) 2 .
[0021] Furthermore, the step S3 is specifically as follows:
[0022] The current prediction equation for the kth cycle is derived based on the current prediction model. The incremental current prediction equation is obtained by taking the difference between the prediction equations of two adjacent cycles:
[0023]
[0024] Let the predicted current be equal to the reference current, and the incremental voltage prediction model is obtained as follows:
[0025]
[0026] Among them, T s is the control period, u d k-1 、u q k-1 are the voltage components of the k-1th cycle of the dq axis, u d k 、u q k are the voltage components of the kth cycle of the dq axis, i d k-1 、i q k-1 are the sampling currents of the k-1th cycle of the dq axis, i d k 、i q k are the k-th cycle sampling currents of dq axis, i d k+1 、i q k+1 are the predicted currents of the k+1th period of the dq axis, L q , L d is the DC-axis inductance, R s is the stator resistance, ω eThe rotor electrical angular velocity is calculated and the flux term is eliminated in the prediction equation, which improves the robustness of the system flux.
[0027] Furthermore, the step S4 is specifically as follows:
[0028] After obtaining the reference voltage in step S3, its vector angle θ ref To perform the calculation, select the angle θ with the vector ref The close vector is used as the first vector of the synthetic reference vector, and the selection range of the second vector is narrowed down to two adjacent vectors of the first vector.
[0029] Furthermore, the step S5 is specifically as follows:
[0030] The duty cycle of the selected voltage vector scheme is allocated according to the Lagrange multiplier method, and the synthesized voltage vector and current prediction value are expressed as:
[0031]
[0032] v hc =d f v f +d s v s
[0033]
[0034] Among them, i d0 k+1 、i df k+1 、i ds k+1 are the d-axis current values after the zero vector, the first vector, and the second vector act alone for one control cycle, i q0 k+1 、i qf k+1 、i qs k+1 are the q-axis current values after the zero vector, the first vector and the second vector act alone for one control cycle, i d * 、i q * are the dq axis reference currents, d0, d f d s are the duty cycles of the zero vector, the first vector, and the second vector respectively, D is the intermediate variable, v hc is the voltage vector synthesized after duty cycle distribution, v f 、v s For the first and second vectors selected, i dhc k+1 with i qhck+1 is the predicted current after the action of the synthetic voltage vector.
[0035] Furthermore, the step S6 is specifically as follows:
[0036] The value function is updated to the difference between the predicted current and the reference current under the action of the synthetic vector, which is shown in the following formula:
[0037] J(v hc )=(i d * -i dhc k+1 ) 2 +(i q * -i qhc k+1 ) 2
[0038] Thus, the global optimal vector solution after duty cycle allocation is obtained.
[0039] Compared with the prior art, the present invention adopts the above technical solution and has the following beneficial effects:
[0040] (1) The present invention proposes a multi-vector model predictive control method based on incremental global optimization, which uses the incremental equation obtained by taking the difference between the periodic current state equations of two adjacent controllers to eliminate the flux term, thereby improving the robustness. The reference voltage vector is obtained according to the incremental voltage prediction equation, thereby reducing the vector control set and reducing the amount of calculation.
[0041] (2) The present invention proposes a multi-vector model predictive control method based on incremental global optimization. After selecting the first vector according to the reference voltage vector, two adjacent vectors are selected as the second voltage vector of the synthetic vector, thereby reducing the candidate vector set and reducing the impact of the reference vector calculation error. The calculation of the value function takes into account the effect of the synthetic vector after the duty cycle is allocated, so that the obtained solution is a global optimization result, which improves the prediction accuracy of the system.
[0042] (3) The present invention proposes a multi-vector model predictive control method based on incremental global optimization. The proposed scheme can be extended to the motor control of N*3 phase structure, with a wide range of applications and strong scalability. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] Figure 1 This is a control framework diagram of the present invention;
[0044] Figure 2 is a vector distribution diagram of the present invention;
[0045] Figure 3 A comparison diagram of the phase current waveforms of the method proposed in this invention and the traditional multi-vector method;
[0046] Figure 4 A comparison diagram of the dq axis current waveforms of the method proposed in the present invention and the traditional multi-vector method;
[0047] Figure 5 This is a comparison diagram of the q-axis current tracking performance of the method proposed in the present invention and the traditional multi-vector method under flux mismatch. DETAILED DESCRIPTION
[0048] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0049] A multi-vector model predictive control method based on incremental global optimization is characterized in that the method controls a permanent magnet synchronous motor through the following steps:
[0050] Step S1, constructing a discretization and current prediction model: using the Euler method to discretize the dq axis current state equations to obtain a current prediction model including dq axis voltage components, quadrature and direct axis inductances, stator resistance, rotor electrical angular velocity, and permanent magnet flux parameters;
[0051] Step S2, setting the value function: using a virtual vector with zero harmonic plane component as the vector control set, and setting the value function to the sum of the squares of the differences between the dq-axis predicted current and the reference current;
[0052] Step S3, deriving an incremental equation and constructing a voltage prediction model: deriving a current prediction equation for the kth cycle, subtracting the prediction current equations of two adjacent cycles to obtain an incremental current prediction equation, setting the predicted current equal to the reference current, and obtaining an incremental voltage prediction model that eliminates the flux linkage term;
[0053] Step S4, vector selection: Calculate the vector angle of the reference voltage, select a vector close to the vector angle according to the vector set as the first vector of the synthetic reference vector, and narrow the selection range of the second vector to two adjacent vectors of the first vector;
[0054] Step S5, duty cycle allocation and synthesis: performing duty cycle allocation on the selected voltage vector scheme according to the Lagrange multiplier method to obtain the synthesized voltage vector and current prediction value after duty cycle allocation;
[0055] Step S6, updating the value function and determining the solution: updating the value function to the difference between the predicted current and the reference current under the action of the synthetic vector, and obtaining the global optimized vector solution after the duty cycle is allocated.
[0056] Based on the above analysis, the multi-vector model predictive current control architecture based on incremental global optimization designed in this patent is as follows Figure 1 shown.
[0057] Furthermore, the step S1 is specifically as follows:
[0058] The current prediction model is obtained by discretizing the dq axis current state equation using the Euler method:
[0059]
[0060] Among them, T s is the control period, u d k 、u q k are the voltage components of the kth cycle of the dq axis, i d k 、i q k are the k-th cycle sampling currents of dq axis, i d k+1 、i q k+1 They are the predicted current of dq axis k+1 period, L q , L d is the DC-axis inductance, R s is the stator resistance, ω e is the rotor electrical angular velocity, is the permanent magnet flux.
[0061] Furthermore, the step S2 is specifically as follows:
[0062] A virtual vector with zero harmonic plane components is used as the vector control set, and the cost function is the sum of the squares of the differences between the dq-axis predicted current and the reference current:
[0063] J=(i d * -i d k+1 ) 2 +(i q * -i q k+1 ) 2 .
[0064] Furthermore, the step S3 is specifically as follows:
[0065] The current prediction equation for the kth cycle is derived based on the current prediction model. The incremental current prediction equation is obtained by taking the difference between the prediction equations of two adjacent cycles:
[0066]
[0067] Let the predicted current be equal to the reference current, and the incremental voltage prediction model is obtained as follows:
[0068]
[0069] Among them, T s is the control period, u d k-1 、u q k-1 are the voltage components of the k-1th cycle of the dq axis, u d k 、u q k are the voltage components of the kth cycle of the dq axis, i d k-1 、i q k-1 are the sampling currents of the k-1th cycle of the dq axis, i d k 、i q k are the k-th cycle sampling currents of dq axis, i d k+1 、i q k+1 are the predicted currents of the k+1th period of the dq axis, L q , L d is the DC-axis inductance, R s is the stator resistance, ω e The rotor electrical angular velocity is calculated and the flux term is eliminated in the prediction equation, which improves the robustness of the system flux.
[0070] Furthermore, the step S4 is specifically as follows:
[0071] After obtaining the reference voltage in step S3, its vector angle θ ref To perform calculations, the vector set is as follows Figure 2 As shown, choose the angle θ with the vector ref The close vector is used as the first vector of the synthetic reference vector, and the selection range of the second vector is narrowed down to two adjacent vectors of the first vector.
[0072] Furthermore, the step S5 is specifically as follows:
[0073] The duty cycle of the selected voltage vector scheme is allocated according to the Lagrange multiplier method, and the synthesized voltage vector and current prediction value are expressed as:
[0074]
[0075] v hc =d f v f+d s v s
[0076]
[0077] Among them, i d0 k+1 、i df k+1 、i ds k+1 are the d-axis current values after the zero vector, the first vector, and the second vector act alone for one control cycle, i q0 k+1 、i qf k+1 、i qs k+1 are the q-axis current values after the zero vector, the first vector and the second vector act alone for one control cycle, i d * 、i q * are the dq axis reference currents, d0, d f d s are the duty cycles of the zero vector, the first vector, and the second vector respectively, D is the intermediate variable, v hc is the voltage vector synthesized after duty cycle distribution, v f 、v s For the first and second vectors selected, i dhc k+1 with i qhc k+1 is the predicted current after the action of the synthetic voltage vector.
[0078] Furthermore, the step S6 is specifically as follows:
[0079] The value function is updated to the difference between the predicted current and the reference current under the action of the synthetic vector, which is shown in the following formula:
[0080] J(v hc )=(i d * -i dhc k+1 ) 2 +(i q * -i qhc k+1 ) 2
[0081] Thus, the global optimal vector solution after duty cycle allocation is obtained.
[0082] Since the voltage vector is predicted without beats, the computational burden is reduced. The computational burden of the proposed method is compared with that of the traditional multi-vector control method and the traditional global optimization method as shown in Table 1. The proposed method reduces the computational burden by 69% compared with the traditional global optimization method.
[0083] Table 1 Comparison of computational complexity between existing methods and the method of the present invention
[0084]
[0085]
[0086] Combine Figure 3 、 4 It can be seen that since the proposed method takes into account the optimal vector solution after duty cycle allocation, it has higher accuracy than the traditional method of first selecting two vectors using the value function and then performing vector synthesis, with smaller dq axis current pulsation and higher phase current sinusoidality. Figure 5 It can be seen that when the controller flux parameters are 50% of the motor flux parameters, there is a static error between the actual q-axis current and the reference current using the traditional method, while the proposed method accurately tracks the given value. Table 2 shows a comparison of the steady-state performance of the proposed method and the traditional multi-vector method. The current ripple calculation method is based on the standard deviation. The proposed method reduces the q-axis ripple by 25.17%, the d-axis ripple by 19.54%, and the phase current THD by 68.61%.
[0087] Table 2 Comparison of steady-state performance between existing methods and the method of the present invention
[0088]
[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A multi-vector model predictive control method based on incremental global optimization, characterized in that: The method realizes control of the permanent magnet synchronous motor by the following steps: Step S1, constructing a discretization and current prediction model: using the Euler method to discretize the dq axis current state equations to obtain a current prediction model including dq axis voltage components, quadrature and direct axis inductances, stator resistance, rotor electrical angular velocity, and permanent magnet flux parameters; Step S2, setting the value function: using a virtual vector with zero harmonic plane component as the vector control set, and setting the value function to the sum of the squares of the differences between the dq-axis predicted current and the reference current; Step S3, deriving an incremental equation and constructing a voltage prediction model: deriving a current prediction equation for the kth cycle, subtracting the prediction current equations of two adjacent cycles to obtain an incremental current prediction equation, setting the predicted current equal to the reference current, and obtaining an incremental voltage prediction model that eliminates the flux linkage term; Step S4, vector selection: Calculate the vector angle of the reference voltage, select a vector close to the vector angle according to the vector set as the first vector of the synthetic reference vector, and narrow the selection range of the second vector to two adjacent vectors of the first vector; Step S5, duty cycle allocation and synthesis: performing duty cycle allocation on the selected voltage vector scheme according to the Lagrange multiplier method to obtain the synthesized voltage vector and current prediction value after duty cycle allocation; Step S6, updating the value function and determining the solution: updating the value function to the difference between the predicted current and the reference current under the action of the synthetic vector, and obtaining the global optimized vector solution after the duty cycle is allocated.
2. The multi-vector model predictive control method based on incremental global optimization according to claim 1, characterized in that: The step S1 is specifically as follows: The current prediction model is obtained by discretizing the dq axis current state equation using the Euler method: Among them, T s is the control period, u d k 、u q k are the voltage components of the kth cycle of the dq axis, i d k 、i q k are the k-th cycle sampling currents of dq axis, i d k+1 、i q k+1 They are the predicted current of dq axis k+1 period, L q 、L d is the DC-axis inductance, R s is the stator resistance, ω e is the rotor electrical angular velocity, is the permanent magnet flux.
3. The multi-vector model predictive control method based on incremental global optimization according to claim 1, characterized in that: The step S2 is specifically as follows: A virtual vector with zero harmonic plane components is used as the vector control set, and the cost function is the sum of the squares of the differences between the dq-axis predicted current and the reference current: J=(i d * -i d k+1 ) 2 +(i q * -i q k+1 ) 2 。 4. The multi-vector model predictive control method based on incremental global optimization according to claim 1, characterized in that: The step S3 is specifically as follows: The current prediction equation for the kth cycle is derived based on the current prediction model. The incremental current prediction equation is obtained by taking the difference between the prediction equations of two adjacent cycles: Let the predicted current be equal to the reference current, and the incremental voltage prediction model is obtained as follows: Among them, T s is the control period, u d k-1 、u q k-1 are the voltage components of the k-1th cycle of the dq axis, u d k 、u q k are the voltage components of the kth cycle of the dq axis, i d k-1 、i q k-1 are the sampling currents of the k-1th cycle of the dq axis, i d k 、i q k are the k-th cycle sampling currents of dq axis, i d k+1 、i q k+1 are the predicted currents of the k+1th period of the dq axis, L q 、L d is the DC-axis inductance, R s is the stator resistance, ω e The rotor electrical angular velocity is calculated and the flux term is eliminated in the prediction equation, which improves the robustness of the system flux.
5. The multi-vector model predictive control method based on incremental global optimization according to claim 1, characterized in that: The step S4 is specifically as follows: After obtaining the reference voltage in step S3, its vector angle θ ref To perform the calculation, select the angle θ with the vector ref The close vector is used as the first vector of the synthetic reference vector, and the selection range of the second vector is narrowed down to two adjacent vectors of the first vector.
6. The multi-vector model predictive control method based on incremental global optimization according to claim 1, characterized in that: The step S5 is specifically as follows: The duty cycle of the selected voltage vector scheme is allocated according to the Lagrange multiplier method, and the synthesized voltage vector and current prediction value are expressed as: v hc =d f v f +d s v s Among them, i d0 k+1 、i df k+1 、i ds k+1 are the d-axis current values after the zero vector, the first vector, and the second vector act alone for one control cycle, i q0 k+1 、i qf k+1 、i qs k+1 are the q-axis current values after the zero vector, the first vector and the second vector act alone for one control cycle, i d * 、i q * are the dq axis reference currents, d0, d f d s are the duty cycles of the zero vector, the first vector, and the second vector respectively, D is the intermediate variable, v hc is the voltage vector synthesized after duty cycle distribution, v f 、v s For the first and second vectors selected, i dhc k+1 with i qhc k+1 is the predicted current after the action of the synthetic voltage vector.
7. The multi-vector model predictive control method based on incremental global optimization according to claim 1, characterized in that: The step S6 is specifically as follows: The value function is updated to the difference between the predicted current and the reference current under the action of the synthetic vector, which is shown in the following formula: J(v hc )=(i d * -i dhc k+1 ) 2 +(i q * -i qhc k+1 ) 2 Thus, the global optimal vector solution after duty cycle allocation is obtained.