Model prediction control method and device for permanent magnet synchronous motor

Abstract

The invention discloses a model prediction control method and device for a permanent magnet synchronous motor, and relates to the technical field of permanent magnet synchronous motor control, and the method comprises the steps: constructing a discrete prediction model based on a mathematical model, and constructing a cost function; converting the cost function into a least square form and introducing an integer variable constraint to obtain a target prediction function, constructing a 3N-layer search tree according to a prediction step length based on the target prediction function, and searching in the first three layers by adopting a full search strategy to obtain a first candidate solution set; from the fourth layer to the 3N-1 layer, after a variable search boundary is set, searching is carried out based on the variable search boundary, and a second candidate solution set is obtained; and on the basis of the first candidate solution set and the second candidate solution set, obtaining a switching sequence of a voltage vector candidate solution which enables the target prediction function to obtain the minimum value at the 3Nth layer, and taking the switching sequence as optimal control output of prediction control. According to the method, the selection of the candidate solution of the optimal solution is ensured, and the calculation burden caused by frequent sorting in the middle process is avoided.

Description

Technical Field

[0001] The present application relates to the technical field of permanent magnet synchronous motor control, and more specifically, to a model predictive control method and device for a permanent magnet synchronous motor. Background Art

[0002] Permanent magnet synchronous motors (PMSMs) have been widely used in industry due to their high efficiency, high power density, and low maintenance costs. To further improve the control performance of PMSMs, model predictive control (MPC) has been introduced as an advanced control strategy. Finite set MPC utilizes rolling optimization and multivariable constraint processing to significantly enhance the dynamic response and robustness of control systems. Research has shown that increasing the prediction step size can improve control performance. However, the computational complexity of multi-step MPC increases exponentially with the prediction step size, affecting its real-time performance and limiting its application in highly dynamic response systems.

[0003] To address the computational overhead associated with multiple steps, the sphere decoding algorithm was introduced into multi-step model predictive control. Traditional sphere decoding algorithms achieve optimal performance while reducing computational overhead by dynamically adjusting the search radius.

[0004] However, selecting the initial search radius for traditional sphere decoding algorithms is complex. The K-best sphere decoding algorithm uses a breadth-first search strategy, using a sorting algorithm to retain only a certain number of candidate solutions at each level. This reduces computational effort and eliminates the need to calculate an initial radius. However, since the sorting results are used at the beginning of the search, there's still a high probability of discarding good candidate solutions. Summary of the Invention

[0005] In response to at least one defect or improvement need in the prior art, the present invention provides a model predictive control method and device for a permanent magnet synchronous motor, which is used to solve the defects of the prior art algorithms in processing multi-step predictions, such as either requiring complex initial parameter settings or discarding potential optimal solutions prematurely during the search process, thereby affecting control performance or computational efficiency.

[0006] To achieve the above objectives, according to a first aspect of the present invention, a model predictive control method for a permanent magnet synchronous motor is provided, comprising:

[0007] Constructing a discrete prediction model based on a preset permanent magnet synchronous motor mathematical model, and constructing a cost function according to the discrete prediction model;

[0008] After converting the cost function into a least squares form and introducing integer variable constraints, a target prediction function is obtained, and a 3N-layer search tree is constructed based on the target prediction function according to the prediction step size, where N is the prediction step size and N ≥ 2;

[0009] At the first three search levels of the search tree, a full search strategy is used to search, retain all candidate solutions of the voltage vector, and obtain a first candidate solution set;

[0010] Dynamically setting a variable search limit based on a probability distribution of the number of candidate voltage vector solutions at each level in the search tree from the fourth level to the 3N-1 level to obtain a second candidate solution set;

[0011] Based on the first candidate solution set and the second candidate solution set, a switching sequence of a voltage vector candidate solution that minimizes the target prediction function is obtained at the 3Nth level of the search tree as the optimal control output of the predictive control.

[0012] Furthermore, in the search levels from the fourth level to the 3N-1 level of the search tree, a variable search limit is dynamically set based on the probability distribution of the number of candidate solutions for the voltage vector at each level, and then a search is performed based on the variable search limit, whereby a second candidate solution set is obtained, including:

[0013] Get the Euclidean distance of the nodes in this layer after the upper layer nodes are expanded;

[0014] Based on the Euclidean distance of the nodes in the current layer, a preset ratio of the number of nodes in the hypersphere is used as the current layer search limit to obtain a variable search limit; the variable search limit includes a maximum number of retained candidates;

[0015] When the number of voltage vector candidate solutions within the hypersphere exceeds the maximum number of retained candidates at the current level, the part exceeding the maximum number of retained candidates is eliminated to obtain the candidate solutions of the next level, so as to complete the search of the fourth to 3N-1 search levels and obtain the second candidate solution set.

[0016] Furthermore, obtaining a switching sequence of a voltage vector candidate solution that minimizes the target prediction function at the 3Nth level of the search tree based on the first candidate solution set and the second candidate solution set as an optimal control output of the predictive control includes:

[0017] A global sorting is performed at the 3Nth search level based on the first candidate solution set and the second candidate solution set, and the switching sequence of the voltage vector candidate solution corresponding to the minimum Euclidean distance is used as the optimal control output of the predictive control.

[0018] Furthermore, after converting the cost function into a least squares form and introducing integer variable constraints, a target prediction function is obtained, including:

[0019] Converting the cost function into a least squares form to obtain a first function;

[0020] Based on the first function, an integer variable constraint is introduced to perform algebraic operations to obtain a second function;

[0021] Perform Cholesky decomposition based on the auxiliary matrix of the second function to obtain a target prediction function.

[0022] Furthermore, the target prediction function is specifically:

[0023]

[0024] Where D is the cumulative Euclidean distance of layer i, d is the cumulative Euclidean distance of the previous i-1 layers, H is the auxiliary function, and S is the switch sequence for multi-step prediction. is the processed unconstrained solution.

[0025] Furthermore, the discrete prediction model is constructed based on the preset permanent magnet synchronous motor mathematical model, including:

[0026] Based on the preset permanent magnet synchronous motor mathematical model, the forward Euler method is used to discretize the continuous-time model to obtain a discrete prediction model.

[0027] Furthermore, the cost function is specifically:

[0028]

[0029] Where, the superscript r represents the reference value, the superscript p represents the predicted value, and λ is the weight coefficient for constraining switching loss.

[0030] According to a second aspect of the present invention, there is also provided a model predictive control device for a permanent magnet synchronous motor, the device comprising:

[0031] A first construction unit is configured to construct a discrete prediction model based on a preset permanent magnet synchronous motor mathematical model, and to construct a cost function according to the discrete prediction model;

[0032] A second construction unit is configured to convert the cost function into a least squares form and introduce integer variable constraints to obtain a target prediction function, and construct a 3N-layer search tree based on the target prediction function according to a prediction step size, where N is the prediction step size and N ≥ 2;

[0033] a first search unit, configured to search the first three search levels of the search tree using a full search strategy, retain all candidate voltage vector solutions, and obtain a first candidate solution set;

[0034] a second search unit, configured to search levels from the fourth level to the 3N-1 level of the search tree, dynamically set a variable search limit based on a probability distribution of the number of candidate solutions for the voltage vector at each level, and then search based on the variable search limit to obtain a second candidate solution set;

[0035] A processing unit is used to obtain a switching sequence of a voltage vector candidate solution that minimizes the target prediction function at the 3Nth layer of the search tree based on the first candidate solution set and the second candidate solution set, as an optimal control output of the predictive control.

[0036] According to a third aspect of the present invention, a computer-readable storage medium is further provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned model predictive control method for the permanent magnet synchronous motor when running.

[0037] According to a fourth aspect of the present invention, an electronic device is also provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the above-mentioned model predictive control method of the permanent magnet synchronous motor through the computer program.

[0038] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0039] The present invention provides a model predictive control method for a permanent magnet synchronous motor. A discrete prediction model is constructed based on a preset permanent magnet synchronous motor mathematical model, and a cost function is constructed based on the discrete prediction model. The cost function is converted into a least squares form and an integer variable constraint is introduced to obtain a target prediction function. A 3N-layer search tree is constructed based on the target prediction function and a prediction step size. A full search strategy is used in the first three search levels of the search tree to retain all candidate voltage vector solutions and obtain a first candidate solution set. From the fourth to the 3N-1 search levels of the search tree, a variable search limit is dynamically set based on the probability distribution of the number of candidate voltage vector solutions at each level, and a search is performed based on the variable search limit to obtain a second candidate solution set. Based on the first and second candidate solution sets, a switching sequence of candidate voltage vector solutions that minimizes the target prediction function is obtained at the 3N level of the search tree, which serves as the optimal control output for predictive control. By using a full search strategy in the first three levels to retain all candidate voltage vector solutions, the present invention avoids the problem of discarding good candidate solutions in the initial stage. Starting from the fourth layer, a variable search bound is dynamically set based on the probability distribution of the number of candidate voltage vector solutions at each layer. At the 3Nth layer, the switching sequence of the candidate voltage vector solution that minimizes the cost function is obtained as the optimal control output for predictive control. By combining a full search with a variable bound search, the comprehensiveness of the search is ensured while controlling computational complexity. Furthermore, by setting a maximum number of valid candidates, a sorting operation is performed only at the last layer, selecting the switching sequence output that minimizes the cost function. This ensures the selection of the optimal solution while avoiding the computational burden of frequent sorting during the intermediate steps. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0041] Figure 1 A flow chart of an optional model predictive control method for a permanent magnet synchronous motor provided in an embodiment of the present application;

[0042] Figure 2 A schematic block diagram of an optional permanent magnet synchronous motor model provided in an embodiment of the present application;

[0043] FIG3 is a schematic block diagram of a voltage vector of an optional three-phase two-level voltage source inverter provided in an embodiment of the present application;

[0044] Figure 4 A schematic block diagram of an optional model predictive current control provided in an embodiment of the present application;

[0045] Figure 5 A schematic flow chart of an optional model predictive control method for a permanent magnet synchronous motor provided in an embodiment of the present application;

[0046] Figure 6 A block diagram of the principle of an optional model predictive control of a permanent magnet synchronous motor provided in an embodiment of the present application;

[0047] Figure 7 A flowchart of a search process for an optional model predictive control of a permanent magnet synchronous motor provided in an embodiment of the present application;

[0048] Figure 8 A schematic block diagram of the structure of an optional electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0049] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0050] The terms "first," "second," "third," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish between different objects, not to describe a particular order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements, but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0051] According to one aspect of the embodiments of the present application, a model predictive control method for a permanent magnet synchronous motor is provided. Figure 1 The present invention describes a model predictive control method for a permanent magnet synchronous motor provided in an embodiment of the present application.

[0052] Figure 1 is a flow chart of an optional model predictive control method for a permanent magnet synchronous motor provided in an embodiment of the present application, such as Figure 1 As shown, the process of the method may include the following steps (STEP, abbreviated as S):

[0053] S101, constructing a discrete prediction model based on a preset permanent magnet synchronous motor mathematical model, and constructing a cost function according to the discrete prediction model;

[0054] S102, converting the cost function into a least squares form and introducing integer variable constraints to obtain a target prediction function, and constructing a 3N-layer search tree based on the target prediction function and the prediction step size;

[0055] S103, searching the first three search levels of the search tree using a full search strategy, retaining all candidate voltage vector solutions to obtain a first candidate solution set;

[0056] S104, searching levels 4 to 3N-1 in the search tree, dynamically setting a variable search limit based on a probability distribution of the number of candidate voltage vector solutions at each level, and then performing a search based on the variable search limit to obtain a second candidate solution set;

[0057] S105 , based on the first candidate solution set and the second candidate solution set, obtaining a switching sequence of the voltage vector candidate solution that makes the target prediction function obtain a minimum value at the 3Nth level of the search tree as the optimal control output of the predictive control.

[0058] An optional model predictive control method for a permanent magnet synchronous motor provided in the present application can be used in the model predictive control of a permanent magnet synchronous motor. It should be noted that the permanent magnet synchronous motor here can be a three-phase permanent magnet synchronous motor or other multi-phase permanent magnet synchronous motor.

[0059] Regarding S101 , a discrete prediction model is constructed based on a preset permanent magnet synchronous motor mathematical model, and a cost function is constructed according to the discrete prediction model.

[0060] In the model predictive control of the embodiment of the present application, taking the multi-step model predictive control of a three-phase permanent magnet synchronous motor as an example, Figure 2 is a schematic block diagram of an optional permanent magnet synchronous motor model provided in an embodiment of the present application, such as Figure 2 As shown in FIG, the permanent magnet synchronous motor consists of a rotor permanent magnet, a rotor core, a stator core and a stator winding.

[0061] In an optional embodiment, a mathematical model of a three-phase permanent magnet synchronous motor may be constructed in the following manner:

[0062] (1) Constructing a model in a three-phase stationary coordinate system

[0063] Without considering factors such as magnetic saturation, the voltage equation of the permanent magnet synchronous motor can be:

[0064]

[0065] Among them, u x (x=a,b,c) is the stator voltage, i x (x=a,b,c) is the stator current, ψ x (x=a,b,c) is the stator winding flux, and R is the stator resistance.

[0066] The magnetic flux equation is:

[0067]

[0068] Among them L xx (x=a,b,c) is the self-inductance of the stator winding, M xy (x, y=a, b, c and x≠y) is the mutual inductance between windings, ψ fx (x = a, b, c) is the cross-linkage flux of the permanent magnet on the stator winding.

[0069] In a three-phase stationary coordinate system, permanent magnet synchronous motors exhibit nonlinear and strongly coupled characteristics, primarily due to the fact that the permanent magnet flux, stator winding self-inductance, and interphase mutual inductance vary with rotor position. This complexity complicates the analysis and control of permanent magnet synchronous motors.

[0070] (2) Constructing the model in the dq axis coordinate system

[0071] To simplify the problem, coordinate transformation methods are usually introduced, including Clarke transformation and Park transformation.

[0072] Clarke transformation converts the three-phase stationary coordinate system into a two-phase stationary coordinate system, and the expression is:

[0073]

[0074] The Park transform further converts it into a dq coordinate system rotating at the synchronous speed of the rotor:

[0075]

[0076] Where θ is the rotor position angle.

[0077] The voltage equation can be expressed in the dq coordinate system as follows:

[0078]

[0079] where u dq is the stator dq axis voltage, i dq is the stator dq axis current, L dq is the stator dq-axis inductance, ω e is the electrical angular velocity, ψ f is the permanent magnet flux linkage.

[0080] The electromagnetic torque equation is:

[0081]

[0082] Where T e is the electromagnetic torque, and p is the number of motor pole pairs.

[0083] The motion equation of the permanent magnet synchronous motor is:

[0084]

[0085] where ω m is the mechanical angular velocity, B is the damping coefficient, J is the moment of inertia, T L is the load torque.

[0086] The permanent magnet synchronous motor is powered by a two-level voltage source inverter with a structure like Figure 2 The state of the inverter bridge arm is shown by s x (x=a,b,c) means, where s i =1 means the upper tube is on and the lower tube is off; s i =0 means the lower tube is open and the upper tube is closed. Therefore, there are 2 3 =8 switching states, each switching state can generate a corresponding voltage vector. FIG3 is a schematic block diagram of an optional three-phase two-level voltage source inverter voltage vector provided in an embodiment of the present application, as shown in FIG3 .

[0087] In an optional embodiment, the model predictive control method for a permanent magnet synchronous motor proposed in this application constructs a discrete prediction model based on a preset permanent magnet synchronous motor mathematical model, including:

[0088] Based on the preset permanent magnet synchronous motor mathematical model, the forward Euler method is used to discretize the continuous-time model to obtain a discrete prediction model.

[0089] Specifically, in each control cycle, the model predictive control traverses all possible switching states of the inverter and their corresponding future system states, and selects the switching state closest to the target state for control. Figure 4 A schematic block diagram of an optional model predictive current control provided in an embodiment of the present application, such as Figure 4 shown.

[0090] Model predictive control uses mathematical models to predict the future behavior of the system based on the switching state of the inverter. To achieve discretization calculation, the forward Euler method is used to discretize equation (5):

[0091]

[0092] Where x=[i d i q ] T is the state variable, y=[i d i q ] T is the system output, s=[s a s b s c ] T is the switching state of the three-phase inverter,

[0093]

[0094] The cost function is used to quantitatively evaluate the closeness between the predicted future state and the target state, so as to select the optimal switching action output.

[0095] Regarding S102, after converting the cost function into a least square form and introducing integer variable constraints, a target prediction function is obtained, and a 3N-layer search tree is constructed based on the target prediction function according to the prediction step size.

[0096] Where N is the prediction step size and N≥2.

[0097] In the multi-step model predictive current control, assuming the prediction range is N, the corresponding cost function can be expressed as:

[0098]

[0099] Where, the superscript r represents the reference value, the superscript p represents the predicted value, and λ is the weight coefficient for constraining switching loss.

[0100] Specifically, as the prediction step increases, the number of voltage vectors increases exponentially. The number of voltage vector combinations for N-step prediction is 8 N , resulting in a significant increase in the computational burden. To this end, this application first transforms the problem into a least squares problem associated with the voltage vector sequence. That is, for the case where there are discrete control variables (such as switch states) in the system, the cost function is transformed into an optimization problem with the goal of minimizing the sum of squares of tracking errors while requiring some variables to be integers. Then, a search tree is constructed, in which each layer corresponds to the switch state of a phase, and every three layers represent a prediction step, so a total of 3N layers are included under the N-step prediction range.

[0101] Regarding S103 , at the first three search levels of the search tree, a full search strategy is used to search, retaining all candidate solutions of the voltage vector to obtain a first candidate solution set;

[0102] At the beginning of the search process, most candidate solutions lie within the sphere, with relatively small Euclidean distances. Therefore, this application employs a full search strategy in the first three levels of the search tree, without limiting the number of candidates at each level. All voltage vector candidate solutions are retained to obtain the first set of candidate solutions. By employing a full search in the first three levels, this application avoids the problem of discarding good candidate solutions in the initial stages.

[0103] Regarding S104, in the search levels from the fourth level to the 3N-1 level of the search tree, a variable search limit is dynamically set based on the probability distribution of the number of candidate solutions for the voltage vector at each level, and then a search is performed based on the variable search limit to obtain a second candidate solution set.

[0104] Specifically, starting from the fourth level, a variable search bound is set based on the probability distribution of the number of candidate solutions within the hypersphere at different search levels. A search is then performed based on this variable search bound to obtain the second set of candidate solutions. This application sets a variable search bound starting from the fourth level. By combining a full search with a variable bound search, the comprehensiveness of the search is ensured while controlling computational complexity.

[0105] As an optional embodiment, in the model predictive control method for a permanent magnet synchronous motor provided by the present invention, in the search levels from the fourth level to the 3N-1 level of the search tree, a variable search limit is dynamically set based on the probability distribution of the number of candidate solutions for the voltage vector at each level, and then a search is performed based on the variable search limit, and the second candidate solution set obtained includes:

[0106] Get the Euclidean distance of the nodes in this layer after the upper layer nodes are expanded;

[0107] Based on the Euclidean distance of the nodes in the current layer, a preset ratio of the number of nodes in the hypersphere is used as the current layer search limit to obtain a variable search limit; the variable search limit includes a maximum number of retained candidates;

[0108] When the number of voltage vector candidate solutions within the hypersphere exceeds the maximum number of retained candidates at the current level, the part exceeding the maximum number of retained candidates is eliminated to obtain the candidate solutions of the next level, so as to complete the search of the fourth to 3N-1 search levels and obtain the second candidate solution set.

[0109] Specifically, the Euclidean distance of the nodes in the current layer obtained after the upper layer nodes are expanded is calculated, and based on the Euclidean distance of the nodes in the current layer, a preset ratio of the number of nodes in the hypersphere is used as the search limit of the current layer to obtain a variable search limit, wherein the variable search limit includes the maximum number of retained candidates. The preset ratio is used as the search limit Max of the current layer i , that is, the maximum number of retained candidates Nmax of this layer is determined i Where k is the current layer number. Max i Equal to p_k times the number of nodes in the hypersphere, p_k is an adjustable parameter. The number of nodes N_k is kept within Max i .

[0110] Regarding S105, based on the first candidate solution set and the second candidate solution set, a switching sequence of the voltage vector candidate solution that makes the target prediction function obtain the minimum value is obtained at the 3Nth level of the search tree as the optimal control output of the predictive control.

[0111] In multi-step model predictive control, computational complexity increases exponentially with increasing prediction step size, impacting the real-time performance of the control system. This method uses a full search in the first three levels of the search tree to obtain a first candidate solution set. Starting from the fourth level, a variable search bound is dynamically set based on the probability distribution of the number of candidate voltage vector solutions at each level. Nodes are screened according to a certain probability and radius to obtain a second candidate solution set. Based on the first and second candidate solution sets, a switching sequence of candidate voltage vector solutions that minimizes the target prediction function is obtained at the 3Nth level of the search tree, serving as the optimal control output for predictive control.

[0112] As an optional embodiment, the model predictive control method for a permanent magnet synchronous motor provided by the present invention obtains, based on the first candidate solution set and the second candidate solution set, a switching sequence of a voltage vector candidate solution that minimizes the target prediction function at the 3Nth level of the search tree as the optimal control output of the predictive control, including:

[0113] A global sorting is performed at the 3Nth search level based on the first candidate solution set and the second candidate solution set, and the switching sequence of the voltage vector candidate solution corresponding to the minimum Euclidean distance is used as the optimal control output of the predictive control.

[0114] As the prediction step increases, the number of voltage vectors increases exponentially. The number of voltage vector combinations for N-step prediction is 8 N , resulting in a significant increase in the computational burden. To this end, in this embodiment, first, the problem is transformed into a least squares problem with integer constraints related to the voltage vector sequence. Then, a search tree is constructed, in which each layer corresponds to the switching state of a phase, and every three layers represent a prediction step. Therefore, a total of 3N layers are included under the N-step prediction range. Next, a global sorting is performed at the 3N search level based on the first and second candidate solution sets, and the switching sequence of the voltage vector candidate solution corresponding to the minimum Euclidean distance is used as the optimal control output of the predictive control, while ensuring optimality and the same control computational complexity. Finally, the optimal voltage vector sequence is output.

[0115] Figure 5 This embodiment provides a flow chart of an optional model predictive control method for a permanent magnet synchronous motor, such as Figure 5 As shown, the motor speed set value ωe* and the actual motor speed ωe are obtained, and the difference between the two is calculated to obtain a speed deviation Δωe; the speed deviation Δωe is input into the PI controller, and the quadrature-axis current set value iq* is obtained through proportional-integral operation, and the direct-axis current set value id* is set to 0; based on the motor mathematical model, combined with the current motor speed ωe and rotor position θ and other operating status information, the prediction model is used to predict the motor direct-axis current and quadrature-axis current idq at the next moment under different switching states;; according to the quadrature-axis current set value iq*, the direct-axis current set value id* and the prediction model output and other information, different voltage vector action sequences are evaluated through the cost function to determine the optimal voltage vector sequence Sopt, and the output is used to control the permanent magnet synchronous motor.

[0116] As an optional embodiment, in the model predictive control method for a permanent magnet synchronous motor provided by the present invention, the cost function is converted into a least squares form and integer variable constraints are introduced to obtain a target prediction function, including:

[0117] Converting the cost function into a least squares form to obtain a first function;

[0118] Based on the first function, an integer variable constraint is introduced to perform algebraic operations to obtain a second function;

[0119] Perform Cholesky decomposition based on the auxiliary matrix of the second function to obtain a target prediction function.

[0120] The first function and the second function represent different forms of the cost function in different calculation processes.

[0121] Specifically, the cost function is converted into a vector form to obtain the first function, which can be shown as follows:

[0122]

[0123] Among them, the output sequence is Y p =[y T (k+1)…y T (k+N)] T =Γx(k)+YS(k)+ZD, the reference value of the output sequence is Auxiliary matrices include Γ, γ, Z, M, E

[0124]

[0125] Based on the algebraic operation of the first function, the second function can be obtained as follows:

[0126] J=(S+Q -1 Θ) T Q(S+Q -1 Θ)+const (11)

[0127] Among them, the auxiliary matrix Θ, Q is defined as (Θ) T =-(Y r -Γx-ZD) T γ-λ(Es(k-1)) T M, Q = γ T γ+λΜ T M.

[0128] The unconstrained solution is:

[0129] S unc =-Q -1 Θ (12)

[0130] Therefore, the second function becomes:

[0131] J=(SS unc ) T Q(SS unc ) (13)

[0132] Based on the auxiliary matrix Q of the second function, Cholesky decomposition is performed to obtain the target prediction function, which specifically includes:

[0133] Perform Cholesky decomposition H on Q T H=Q.

[0134] The second function can be expressed as:

[0135]

[0136] Among them, the processed unconstrained solution is

[0137] Finally, the optimization problem is transformed into a least squares problem, which can be expressed as:

[0138]

[0139] Since H is a lower triangular matrix, the Euclidean distance in Equation (15) can be written as:

[0140]

[0141] Then, the accumulated target prediction function can be expressed as:

[0142]

[0143] Where D is the cumulative Euclidean distance of layer i, d is the cumulative Euclidean distance of the previous i-1 layers, H is the auxiliary function, and S is the switch sequence for multi-step prediction. is the processed unconstrained solution.

[0144] The present invention adopts this mathematical transformation to transform the originally complex multi-step model predictive control problem into a standard least squares problem with integer constraints, providing a theoretical basis for the subsequent use of the sphere decoding algorithm and also improving the efficiency of problem solving.

[0145] The following describes the model predictive control method for a permanent magnet synchronous motor proposed by the present invention using a specific embodiment.

[0146] Traditional sphere decoding algorithms require selecting an appropriate initial radius. A radius that is too small may not contain the optimal solution, while a radius that is too large may reduce computational efficiency. The K-best sphere decoding algorithm has a certain probability of discarding good candidate solutions at the beginning of the search process, and the sorting and comparison at each layer increases the computational burden.

[0147] To solve these problems, the present invention proposes a model predictive control method for a permanent magnet synchronous motor:

[0148] A discrete prediction model is constructed based on a preset mathematical model of a permanent magnet synchronous motor, and a cost function is constructed based on the discrete prediction model. After converting the cost function into a least-squares problem with integer constraints, a 3N-layer search tree is constructed based on the prediction step size, where N is the prediction step size and N ≥ 2.

[0149] Figure 6 The principle block diagram of an optional model predictive control of a permanent magnet synchronous motor provided in this embodiment is as follows: Figure 6As shown, full expansion is used in the first three levels of the search tree. Starting from the fourth level, variable search bounds are set according to the probability distribution of the number of candidate solutions within the hypersphere (determined by the radius) at different search levels, and there is no need to sort the candidate solutions. Once the predetermined maximum number of valid candidates Max is reached, the search tree is searched. i , candidate solutions that exceed the predetermined limit will be discarded from the search process, and the search process will continue at the i+1th level based on the previously found candidate solutions.

[0150] exist Figure 6 In , the first three layers use full search, and subsequent layers use variable search bounds.

[0151] At the beginning of the search process, most candidate solutions lie within the sphere, with relatively small Euclidean distances. This leads to a high error rate when filtering candidate solutions based on the sorted results. This problem can be avoided by using a full search in the first three levels, without limiting the number of candidates at each level.

[0152] Figure 7 A search flow chart of an optional model predictive control of a permanent magnet synchronous motor provided in this embodiment is as follows: Figure 7 As shown, starting from the fourth layer, the Euclidean distance of the nodes in this layer obtained after the upper layer nodes are expanded is first calculated, and then the preset ratio, such as 10 / 11 of the number of nodes in the hypersphere, is used as the search limit Max of this layer. i , that is, the maximum number of retained candidates Nmax of this layer is determined i In the search of the fourth level and later, if the maximum number of valid candidates Nmax is reached i Candidate solutions exceeding the preset ratio are discarded. Upon reaching the 3Nth level, all candidate voltage vector solutions are sorted, and the switching sequence output of the voltage vector candidate solution that minimizes the cost function (that is, the Euclidean distance) is selected. It is important to note that the screening criterion—the preset ratio—can vary depending on the specific application scenario. By studying how control performance varies with the hypersphere radius and screening ratio, we can select an appropriate hypersphere radius and screening ratio as the preset ratio.

[0153] At the beginning of the search process, the present invention has most candidate solutions located within the sphere, and their Euclidean distances are relatively small. By adopting full search in the first three layers, the problem of discarding good candidate solutions in the initial stage is avoided. Starting from the fourth layer, the present invention sets a variable search limit. By combining full search with variable limit search, the comprehensiveness of the search is guaranteed and the computational complexity is controlled. By setting the maximum number of valid candidates, the present invention effectively controls the size of the search space and avoids the exponential growth of computational complexity. The present invention performs sorting operations only when the last layer is searched, and selects the switch sequence output with the smallest Euclidean distance, which not only ensures the selection of the optimal solution, but also avoids the computational burden caused by frequent sorting in the intermediate process. By allowing the adjustment of the hypersphere radius and the screening ratio, the present invention allows the algorithm to be optimized according to the specific application scenario, achieving a balance between control performance and computational efficiency.

[0154] According to another aspect of the embodiments of the present application, a model predictive control device for a permanent magnet synchronous motor is provided for implementing the above-mentioned model predictive control method for a permanent magnet synchronous motor. The device may include:

[0155] A first construction unit is configured to construct a discrete prediction model based on a preset permanent magnet synchronous motor mathematical model, and to construct a cost function according to the discrete prediction model;

[0156] A second construction unit is configured to convert the cost function into a least squares form and introduce integer variable constraints to obtain a target prediction function, and construct a 3N-layer search tree based on the target prediction function according to a prediction step size, where N is the prediction step size and N ≥ 2;

[0157] a first search unit, configured to search the first three search levels of the search tree using a full search strategy, retain all candidate voltage vector solutions, and obtain a first candidate solution set;

[0158] a second search unit, configured to search levels from the fourth level to the 3N-1 level of the search tree, dynamically set a variable search limit based on a probability distribution of the number of candidate solutions for the voltage vector at each level, and then search based on the variable search limit to obtain a second candidate solution set;

[0159] A processing unit is used to obtain a switching sequence of a voltage vector candidate solution that minimizes the target prediction function at the 3Nth layer of the search tree based on the first candidate solution set and the second candidate solution set, as an optimal control output of the predictive control.

[0160] It should be noted here that the examples and scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the contents disclosed in the above embodiments. It should be noted that the above modules as part of the device can run in a hardware environment, can be implemented by software, and can also be implemented by hardware, where the hardware environment includes a network environment.

[0161] According to another aspect of the embodiments of the present application, a storage medium is further provided. Optionally, in this embodiment, the storage medium can be used to execute the program code of any of the above-mentioned permanent magnet synchronous motor model predictive control methods in the embodiments of the present application.

[0162] Optionally, in this embodiment, the storage medium is configured to store program codes for executing the following steps:

[0163] S1: constructing a discrete prediction model based on a preset permanent magnet synchronous motor mathematical model, and constructing a cost function according to the discrete prediction model;

[0164] S2: After converting the cost function into a least squares form and introducing integer variable constraints, a target prediction function is obtained, and a 3N-layer search tree is constructed based on the target prediction function according to the prediction step size, where N is the prediction step size and N≥2;

[0165] S3: searching the first three search levels of the search tree using a full search strategy, retaining all candidate voltage vector solutions, and obtaining a first candidate solution set;

[0166] S4: searching the levels from the fourth level to the 3N-1 level of the search tree, dynamically setting a variable search limit based on the probability distribution of the number of candidate solutions for the voltage vector at each level, and then searching based on the variable search limit to obtain a second candidate solution set;

[0167] S5: Based on the first candidate solution set and the second candidate solution set, a switching sequence of the voltage vector candidate solution that makes the target prediction function achieve the minimum value is obtained at the 3Nth layer of the search tree as the optimal control output of the predictive control.

[0168] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments, which will not be described in detail in this embodiment.

[0169] Among them, computer-readable storage media may include, but are not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic or optical cards, nanosystems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.

[0170] According to another aspect of the embodiments of the present application, an electronic device for implementing the above-mentioned model predictive control method of the permanent magnet synchronous motor is also provided. The electronic device can be a server, a terminal, or a combination thereof.

[0171] Figure 8 is a schematic structural diagram of an optional electronic device according to an embodiment of the present application, such as Figure 8 As shown, it includes a processor 802, a communication interface 804, a memory 806 and a communication bus 808, wherein the processor 802, the communication interface 804, and the memory 806 communicate with each other via the communication bus 808, wherein,

[0172] Memory 806, for storing computer programs;

[0173] The processor 802 is configured to execute the computer program stored in the memory 806 to implement the following steps:

[0174] S1: constructing a discrete prediction model based on a preset permanent magnet synchronous motor mathematical model, and constructing a cost function according to the discrete prediction model;

[0175] S2: After converting the cost function into a least squares form and introducing integer variable constraints, a target prediction function is obtained, and a 3N-layer search tree is constructed based on the target prediction function according to the prediction step size, where N is the prediction step size and N≥2;

[0176] S3: searching the first three search levels of the search tree using a full search strategy, retaining all candidate voltage vector solutions, and obtaining a first candidate solution set;

[0177] S4: searching the search levels from the fourth level to the 3N minus 1 level of the search tree, dynamically setting a variable search limit based on the probability distribution of the number of candidate voltage vector solutions at each level, and then searching based on the variable search limit to obtain a second candidate solution set;

[0178] S5: Based on the first candidate solution set and the second candidate solution set, a switching sequence of the voltage vector candidate solution that makes the target prediction function achieve the minimum value is obtained at the 3Nth layer of the search tree as the optimal control output of the predictive control.

[0179] Optionally, the communication bus may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The communication bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8The communication interface is used for communication between the electronic device and other devices.

[0180] The memory may include RAM, or may include non-volatile memory, such as at least one disk memory. Alternatively, the memory may also be at least one storage device located away from the aforementioned processor.

[0181] The above-mentioned processor can be a general-purpose processor, including but not limited to: CPU (Central Processing Unit), NP (Network Processor), etc.; it can also be DSP (Digital Signal Processing), ASIC (Application Specific Integrated Circuit), FPGA (Field-Programmable Gate Array) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0182] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments, and this embodiment will not be described in detail here.

[0183] It should be noted that for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0184] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0185] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interface, and the indirect coupling or communication connection of the device or unit can be electrical or other forms.

[0186] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0187] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0188] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, including a number of instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned memory includes: various media that can store program codes, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0189] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable memory, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0190] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the specification and practicing the disclosure herein, those skilled in the art will easily think of the implementation scheme of the present disclosure. This application is intended to cover any variation, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the art that are not recorded in the present disclosure. The description and examples are to be regarded as exemplary only, and the scope and spirit of the present disclosure are defined by the claims.

[0191] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0192] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A model predictive control method for a permanent magnet synchronous motor, characterized in that: include: Constructing a discrete prediction model based on a preset permanent magnet synchronous motor mathematical model, and constructing a cost function according to the discrete prediction model; After converting the cost function into a least squares form and introducing integer variable constraints, a target prediction function is obtained, and a 3N-layer search tree is constructed based on the target prediction function according to the prediction step size, where N is the prediction step size and N ≥ 2; At the first three search levels of the search tree, a full search strategy is used to search, retain all candidate solutions of the voltage vector, and obtain a first candidate solution set; Dynamically setting a variable search limit based on a probability distribution of the number of candidate voltage vector solutions at each level in the search tree from the fourth level to the 3N-1 level to obtain a second candidate solution set; Based on the first candidate solution set and the second candidate solution set, a switching sequence of a voltage vector candidate solution that minimizes the target prediction function is obtained at the 3Nth level of the search tree as the optimal control output of the predictive control.

2. The model predictive control method for a permanent magnet synchronous motor according to claim 1, wherein: The search levels from the fourth level to the 3N-1 level of the search tree are searched, and a variable search limit is dynamically set based on the probability distribution of the number of candidate solutions for the voltage vector at each level, and then the search is performed based on the variable search limit to obtain a second candidate solution set including: Get the Euclidean distance of the nodes in this layer after the upper layer nodes are expanded; Based on the Euclidean distance of the nodes in the current layer, a preset ratio of the number of nodes in the hypersphere is used as the current layer search limit to obtain a variable search limit; the variable search limit includes a maximum number of retained candidates; When the number of voltage vector candidate solutions within the hypersphere exceeds the maximum number of retained candidates at the current level, the part exceeding the maximum number of retained candidates is eliminated to obtain the candidate solutions of the next level, so as to complete the search of the fourth to 3N-1 search levels and obtain the second candidate solution set.

3. The model predictive control method for a permanent magnet synchronous motor according to claim 2, wherein: Obtaining, based on the first candidate solution set and the second candidate solution set, a switching sequence of a voltage vector candidate solution that minimizes the target prediction function at the 3Nth level of the search tree as an optimal control output of the predictive control, including: A global sorting is performed at the 3Nth search level based on the first candidate solution set and the second candidate solution set, and the switching sequence of the voltage vector candidate solution corresponding to the minimum Euclidean distance is used as the optimal control output of the predictive control.

4. The model predictive control method for a permanent magnet synchronous motor according to claim 1, wherein: After converting the cost function into the least square form and introducing integer variable constraints, the target prediction function is obtained, including: Converting the cost function into a least squares form to obtain a first function; Based on the first function, an integer variable constraint is introduced to perform algebraic operations to obtain a second function; Perform Cholesky decomposition based on the auxiliary matrix of the second function to obtain a target prediction function.

5. The model predictive control method for a permanent magnet synchronous motor according to claim 4, wherein: The target prediction function is specifically: Where D is the cumulative Euclidean distance of layer i, d is the cumulative Euclidean distance of the previous i-1 layers, H is the auxiliary function, and S is the switch sequence for multi-step prediction. is the processed unconstrained solution.

6. The model predictive control method for a permanent magnet synchronous motor according to claim 1, wherein: The discrete prediction model is constructed based on a preset permanent magnet synchronous motor mathematical model, including: Based on the preset permanent magnet synchronous motor mathematical model, the forward Euler method is used to discretize the continuous-time model to obtain a discrete prediction model.

7. The model predictive control method for a permanent magnet synchronous motor according to claim 1, wherein: The cost function is specifically: Where, the superscript r represents the reference value, the superscript p represents the predicted value, and λ is the weight coefficient for constraining switching loss.

8. A model predictive control device for a permanent magnet synchronous motor, characterized in that: include: A first construction unit is configured to construct a discrete prediction model based on a preset permanent magnet synchronous motor mathematical model, and to construct a cost function according to the discrete prediction model; A second construction unit is configured to convert the cost function into a least squares form and introduce integer variable constraints to obtain a target prediction function, and construct a 3N-layer search tree based on the target prediction function according to a prediction step size, where N is the prediction step size and N ≥ 2; a first search unit, configured to search the first three search levels of the search tree using a full search strategy, retain all candidate voltage vector solutions, and obtain a first candidate solution set; a second search unit, configured to search levels from the fourth level to the 3N-1 level of the search tree, dynamically set a variable search limit based on a probability distribution of the number of candidate solutions for the voltage vector at each level, and then search based on the variable search limit to obtain a second candidate solution set; A processing unit is used to obtain a switching sequence of a voltage vector candidate solution that minimizes the target prediction function at the 3Nth layer of the search tree based on the first candidate solution set and the second candidate solution set, as an optimal control output of the predictive control.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program executes the method according to any one of claims 1 to 7 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 7 through the computer program.

Images