A permanent magnet synchronous motor current prediction control method based on dynamic trajectory optimization

Abstract

The application discloses a permanent magnet synchronous motor current prediction control method based on dynamic trajectory optimization. d‑ q The method collects the current operation state of the motor and obtains target current or target torque, establishes a current discrete prediction model in a coordinate system, determines one-step current reachable domain according to inverter voltage constraint, and discretizes the current plane and the electric angle phase to construct an extended state directed graph. d-q According to a preset optimization target, an optimal path from the current state to the target state is searched, a multi-step current reference trajectory is generated, control voltage is calculated based on the prediction model, and the inverter is driven. The application can optimize the current transition path under the condition of voltage limitation and field weakening, improve the current dynamic response and torque establishment performance, and reduce the transition process loss to a certain extent, and has certain engineering application value.

Description

Technical Field

[0001] This invention relates to current control technology for permanent magnet synchronous motors, and more particularly to a predictive control method for the current of permanent magnet synchronous motors based on dynamic trajectory optimization under voltage-constrained conditions, belonging to the field of power electronics and motor drive control technology. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) are characterized by high power density, high efficiency, and excellent dynamic performance, and have been widely used in new energy vehicles, electric drives, industrial servo systems, and high-end equipment drives. Under conditions such as rapid start-stop, sudden torque changes, high-speed re-acceleration, and field weakening operation, the drive system requires rapid establishment of current and electromagnetic torque to meet high dynamic control demands.

[0003] Existing current control methods include vector control based on proportional-integral regulators and predictive current control. Predictive current control uses a discrete model of the motor to predict future current states and calculates the control input accordingly. It features clear structure, fast response, and ease of digital implementation. Among these, deadbeat predictive current control is often used to achieve fast current tracking.

[0004] However, under conditions of limited DC bus voltage, restricted inverter modulation boundaries, and increased back EMF during high-speed operation, the reference current often cannot be reached directly within a single control cycle. Traditional predictive current control typically calculates the control voltage based primarily on single-step current error. When the reference current is unreachable in one step, the lack of overall organization of the intermediate target current point and subsequent multi-step transition paths may lead to decreased current propulsion efficiency, delayed torque build-up, and increased transition losses under constrained operating conditions.

[0005] Therefore, it is necessary to propose a current predictive control method that can construct a one-step current reachability relationship, coordinate the multi-step current transient process, and take into account response speed, torque build-up, and loss indicators under voltage constraints. Summary of the Invention

[0006] Technical Problem: The purpose of this invention is to provide a current prediction control method for permanent magnet synchronous motors based on dynamic trajectory optimization, which solves the problems that the reference current is difficult to reach directly in a single cycle under voltage-limited conditions and that traditional predictive current control lacks multi-step path organization capabilities, so as to improve the dynamic current propulsion capability and torque build-up performance under limited operating conditions and reduce loss indicators during the transition process.

[0007] Technical Solution: To achieve the above objectives, the present invention provides a current prediction control method for permanent magnet synchronous motors based on dynamic trajectory optimization, which employs the following steps:

[0008] Step S1: Collect the current operating status parameters of the permanent magnet synchronous motor. The operating status parameters include at least the d-axis current, q-axis current, electrical angle, electrical angular velocity, and DC bus voltage, and obtain the target current or target torque.

[0009] Step S2: Based on the operating state parameters, establish a discrete current prediction model for the permanent magnet synchronous motor in the dq coordinate system;

[0010] Step S3: Determine the one-step current reachable domain of the current state within a single control cycle based on the inverter output voltage constraint, and establish a state transition relationship that satisfies the voltage constraint based on the one-step current reachable domain.

[0011] Step S4: Discretize the dq current plane and electrical angle phase to construct an extended state directed graph consisting of discrete current states, electrical angle phases, and the state transition relationships.

[0012] Step S5: Based on the preset optimization objective, search the extended state directed graph for the optimal path from the current state to the target state to obtain the multi-step current reference trajectory.

[0013] Step S6: In each control cycle, select the target current point corresponding to the multi-step current reference trajectory, calculate the control voltage based on the current discrete prediction model, and drive the inverter to realize the current control of the permanent magnet synchronous motor.

[0014] The discrete current prediction model is expressed as follows:

[0015] ,

[0016] in, This is the current dq-axis current vector. This is the current dq-axis voltage vector. , The matrix is ​​determined by the motor parameters, sampling period, and electrical angular velocity. It is a perturbation vector that includes the back potential term.

[0017] The one-step current reachable domain is obtained by mapping the inverter voltage feasible set through the current discrete prediction model; the inverter voltage feasible set is a convex set formed by combining each effective voltage vector and zero vector according to the duty cycle; the one-step current reachable domain is determined by the convex hull composed of multiple predicted current vertices.

[0018] The state transition relationship is determined by the state change that satisfies the one-step current reachability constraint: if the discrete state node of the next control cycle is located in the one-step current reachability domain corresponding to the current state, then a directed edge is established from the current state node to the discrete state node.

[0019] The state nodes in the extended state directed graph include discrete d-axis current, discrete q-axis current, and discrete electrical angle phase; the state transition relationship simultaneously satisfies voltage constraints, discrete mesh mapping constraints, and reachability correction constraints.

[0020] The preset optimization objectives include at least one of the following: the minimum number of control cycles required for the current to reach the target state, the minimum torque gap area, and the minimum transient loss index.

[0021] When the optimization objective is to minimize the number of control cycles, the path cost of each state transition edge is set to a constant; when the optimization objective is to minimize the torque gap area, the path cost of each state transition edge is determined based on the torque gap increment; when the optimization objective is to minimize the transient process loss index, the path cost of each state transition edge is determined based on the single-step loss increment.

[0022] The optimal path search employs a graph search method; specifically, for equally weighted path costs, a layer-by-layer expansion search is used, and for non-negatively weighted path costs, a shortest path search based on cumulative cost updates is used.

[0023] The control voltage is obtained by back-calculating the target current point of the next control cycle and the current state through the current discrete prediction model, and the inverter switching signal is generated by space vector pulse width modulation.

[0024] The multi-step current reference trajectory is pre-generated and stored as a trajectory library offline. During operation, the corresponding trajectory is called according to the current state, target state, and operating condition information; or the discrete granularity of the current state space is adjusted according to the operating conditions to balance trajectory planning accuracy and computational complexity.

[0025] Beneficial effects:

[0026] (1) This invention transforms the current control problem under voltage-limited conditions into a dynamic trajectory optimization problem in a constrained state space, which can organize the current transition path from the perspective of the whole process.

[0027] (2) By constructing a one-step current reachable domain and a state transition relationship that satisfies voltage constraints, the present invention makes the planned trajectory match the actual output capability of the inverter, thereby improving the executability of trajectory planning.

[0028] (3) By setting different path costs, the present invention can perform trajectory planning for rapid current arrival, rapid torque establishment and transient loss suppression respectively, adapting to different dynamic control objectives.

[0029] (4) The present invention adopts a combination of planning layer and execution layer. The execution layer can still use predictive current control to calculate voltage. The structure is clear and easy to implement in engineering.

[0030] (5) The present invention can reduce the online calculation burden by using an offline trajectory library call method, and is suitable for high-performance control of permanent magnet synchronous motors under limited working conditions such as low bus voltage, high speed and weak field. Attached Figure Description

[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the embodiments are briefly described below:

[0032] Figure 1 This is a schematic diagram of the current prediction control method for permanent magnet synchronous motors based on dynamic trajectory optimization as described in this invention. Detailed Implementation

[0033] The present invention will be further described in detail below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the following embodiments.

[0034] The present invention provides a current prediction control method for a permanent magnet synchronous motor based on dynamic trajectory optimization, comprising the following steps:

[0035] Step S1: Collect the current operating status parameters of the permanent magnet synchronous motor, including at least the d-axis current, q-axis current, electrical angle, electrical angular velocity, and DC bus voltage, and obtain the target current or target torque.

[0036] Step S2: Establish a discrete current prediction model based on the discrete dynamic relationship of the permanent magnet synchronous motor in the dq coordinate system. The model is used to describe the mapping relationship between the current state, control voltage and the current state of the next control cycle.

[0037] Step S3: Based on the inverter output voltage constraint, obtain the set of currents that can be reached in a single control cycle under the current state, forming a one-step current reachable domain; establish a state transition relationship that satisfies the voltage constraint based on the one-step current reachable domain.

[0038] Step S4: Discretize the dq current plane and the electrical angle phase to construct an extended state directed graph; the nodes in the directed graph represent discrete current states and phase states, and the edges represent state transitions that satisfy the one-step reachability constraint.

[0039] Step S5: Set the path cost according to the preset optimization objective, and search for the optimal path from the current state to the target state in the extended state directed graph to obtain the multi-step current reference trajectory. The optimization objective includes at least one of the following: minimum number of control cycles, minimum torque gap area, or minimum transition loss.

[0040] Step S6: In each control cycle, the current target node of the multi-step current reference trajectory is called, the control voltage is calculated based on the current prediction model, and the corresponding voltage is output through the inverter, so that the motor current gradually approaches the target state along the optimized trajectory.

[0041] (1) System Overview

[0042] like Figure 1 As shown, the method described in this embodiment is applied to a motor drive system consisting of a permanent magnet synchronous motor, a three-phase inverter, and a controller. The controller is used to complete state acquisition, reachability domain construction, trajectory planning, and control voltage calculation; the inverter is used to generate drive voltage based on the modulation signal output by the controller.

[0043] (2) Status acquisition

[0044] During the k-th control cycle, collect the current operating status parameters, including the d-axis current. q-axis current electrical angle Electric angular velocity and DC bus voltage Depending on the application requirements, the target current can also be obtained. or target torque .

[0045] (3) Establishment of current prediction model

[0046] In the synchronously rotating dq coordinate system, a discrete prediction model is established based on the voltage equation of the permanent magnet synchronous motor, and its vector form can be expressed as:

[0047] ,

[0048] in, For the current vector, To control the voltage vector, , Determined by motor parameters, sampling period, and electrical angular velocity. This is a disturbance vector containing a back EMF term. This model is used to describe the change in the current state in the next cycle after the control voltage is applied.

[0049] (4) Construction of the one-step current reachable domain and state transition relationship

[0050] Within a single control cycle, current variation is limited by the inverter's output voltage capability. Let... Let be the set of voltage feasible values ​​that the inverter can achieve within the current control cycle, then any After being mapped by the current discrete prediction model, each corresponds to a current state in the next cycle.

[0051] Preferably, the feasible set of inverter voltages is determined by a convex set consisting of multiple effective voltage vectors and zero vectors. Mapping this set onto the dq current plane yields the one-step current reachable domain corresponding to the current state. The reachable region can be represented as a convex hull formed by multiple predicted current vertices:

[0052] ,

[0053] When the target current is within the reachable range of the one-step current, it can be directly used as the target for the next cycle; when the target current is outside the reachable range of the one-step current, the transition path for the next few control cycles needs to be determined through trajectory planning.

[0054] To construct a searchable state space, this embodiment discretizes the dq current plane and electrical angle phase. If a discrete node is located within the reachable domain of the current step corresponding to the current state, a state transition edge is established from the current node to that node. This forms a voltage-constrained extended state directed graph.

[0055] (5) Generation of multi-step current trajectory

[0056] In the extended state directed graph, path costs are set according to different control objectives, and the optimal path from the current state to the target state is searched.

[0057] In one implementation, with the goal of minimizing the number of control cycles, each state transition edge is assigned the same cost, and a layer-by-layer expansion search is used to obtain the shortest path to the target neighborhood.

[0058] In another implementation, with the minimum torque gap area as the objective, the path cost is set according to the torque gap increment corresponding to the current edge, and the target trajectory is obtained by minimizing the cumulative path cost.

[0059] In another implementation, with the goal of minimizing transition loss, edge weights are set according to the single-step loss index, and the target trajectory is obtained by minimizing cumulative loss.

[0060] The path search can be implemented using breadth-first search, shortest path search based on cumulative path cost updates, or other graph search algorithms capable of finding the optimal path in a finite state space. The resulting multi-step current reference trajectory is then obtained.

[0061] ,

[0062] Where N is the trajectory length. This is the reference current node corresponding to the (k+n)th control cycle.

[0063] (6) Trajectory execution and control output

[0064] After obtaining the multi-step current reference trajectory, the controller selects the corresponding target current point in each control cycle. The required control voltage is then calculated using a discrete current prediction model.

[0065] ,

[0066] Subsequently, the controller processes the control voltage through the modulation module and drives the inverter, causing the motor current to gradually approach the target state along the planned trajectory. By generating a reference trajectory at the planning layer and tracking it step by step at the execution layer, both trajectory optimization capability and control implementation clarity can be balanced.

[0067] (7) Optional implementation methods

[0068] In some embodiments, the multi-step current reference trajectory can be pre-calculated offline, and a trajectory library can be established based on bus voltage, rotational speed, current state, and target state; during operation, the corresponding trajectory can be called according to the actual working conditions to reduce the online calculation burden.

[0069] In other embodiments, the discrete granularity of the current state space and electrical angle phase can be adjusted according to the operating conditions to strike a trade-off between trajectory planning accuracy and computational complexity.

[0070] Without departing from the spirit and essence of this invention, those skilled in the art may make equivalent substitutions or modifications to the above embodiments, all of which shall fall within the protection scope of this invention.

Claims

1. A current prediction control method for a permanent magnet synchronous motor based on dynamic trajectory optimization, characterized in that, Includes the following steps: Step S1: Collect the current operating status parameters of the permanent magnet synchronous motor. The operating status parameters include at least the d-axis current, q-axis current, electrical angle, electrical angular velocity, and DC bus voltage, and obtain the target current or target torque. Step S2: Based on the operating state parameters, establish a discrete current prediction model for the permanent magnet synchronous motor in the dq coordinate system; Step S3: Determine the one-step current reachable domain of the current state within a single control cycle based on the inverter output voltage constraint, and establish a state transition relationship that satisfies the voltage constraint based on the one-step current reachable domain. Step S4: Discretize the dq current plane and electrical angle phase to construct an extended state directed graph consisting of discrete current states, electrical angle phases, and the state transition relationships. Step S5: Based on the preset optimization objective, search the extended state directed graph for the optimal path from the current state to the target state to obtain the multi-step current reference trajectory. Step S6: In each control cycle, select the target current point corresponding to the multi-step current reference trajectory, calculate the control voltage based on the current discrete prediction model, and drive the inverter to realize the current control of the permanent magnet synchronous motor.

2. The control method according to claim 1, characterized in that, The discrete current prediction model is expressed as follows: ， in, This is the current dq-axis current vector. This is the current dq-axis voltage vector. , The matrix is ​​determined by the motor parameters, sampling period, and electrical angular velocity. It is a perturbation vector that includes the back potential term.

3. The control method according to claim 1, characterized in that, The one-step current reachable domain is obtained by mapping the inverter voltage feasible set through the current discrete prediction model; the inverter voltage feasible set is a convex set formed by combining each effective voltage vector and zero vector according to the duty cycle; the one-step current reachable domain is determined by the convex hull composed of multiple predicted current vertices.

4. The control method according to claim 1, characterized in that, The state transition relationship is determined by the state change that satisfies the one-step current reachability constraint: if the discrete state node of the next control cycle is located in the one-step current reachability domain corresponding to the current state, then a directed edge is established from the current state node to the discrete state node.

5. The control method according to claim 1, characterized in that, The state nodes in the extended state directed graph include discrete d-axis current, discrete q-axis current, and discrete electrical angle phase; the state transition relationship simultaneously satisfies voltage constraints, discrete mesh mapping constraints, and reachability correction constraints.

6. The control method according to claim 1, characterized in that, The preset optimization objectives include at least one of the following: the minimum number of control cycles required for the current to reach the target state, the minimum torque gap area, and the minimum transient loss index.

7. The control method according to claim 6, characterized in that: When the optimization objective is to minimize the number of control cycles, the path cost of each state transition edge is set to a constant; when the optimization objective is to minimize the torque gap area, the path cost of each state transition edge is determined based on the torque gap increment; when the optimization objective is to minimize the transient process loss index, the path cost of each state transition edge is determined based on the single-step loss increment.

8. The control method according to claim 7, characterized in that, The optimal path search employs a graph search method; specifically, for equally weighted path costs, a layer-by-layer expansion search is used, and for non-negatively weighted path costs, a shortest path search based on cumulative cost updates is used.

9. The control method according to claim 1, characterized in that, The control voltage is obtained by back-calculating the target current point of the next control cycle and the current state through the current discrete prediction model, and the inverter switching signal is generated by space vector pulse width modulation.

10. The control method according to claim 1, characterized in that, The multi-step current reference trajectory is pre-generated and stored as a trajectory library offline. During operation, the corresponding trajectory is called according to the current state, target state and operating condition information. Alternatively, the discrete granularity of the current state space can be adjusted according to the operating conditions to balance the accuracy of trajectory planning and computational complexity.

Images