Method for designing vector decoupling angle of open-winding permanent magnet synchronous motor

Abstract

The invention belongs to the technical field of motor control, and provides an open-winding permanent magnet synchronous motor vector decoupling angle design method for solving the problem that current research mostly focuses on static analysis of a fixed decoupling angle, which comprises the following steps: constructing an open-winding permanent magnet synchronous motor mathematical model and a cost function; dividing a training set and a verification set by taking the torque error, the total loss of the motor and the total harmonic current distortion rate of the current as a first data set; optimizing the weight and the bias coefficient of the feedforward neural network; training the optimized feedforward neural network through the training set, and verifying through the verification set to obtain a trained neural network model; taking different working conditions and vector decoupling angles as a second data set, inputting the second data set into the trained neural network model, and outputting prediction performance index data; constructing a database; and dynamic optimization of the open-winding permanent magnet synchronous motor vector decoupling angle under different working conditions is realized through a table look-up method.

Description

Technical Field

[0001] The present invention belongs to the technical field of motor control, and in particular relates to a method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor. Background Art

[0002] The open-winding permanent magnet synchronous motor (PMSM) is a special type of PMSM structure, featuring an open stator winding connection, rather than the traditional closed star or delta connection. This structure provides unique control flexibility and performance advantages, and has garnered widespread attention in recent years across multiple fields. Open-winding PMSMs can be used in electric and hybrid vehicle drive systems, as well as in energy recovery systems for hybrid vehicles. They feature high power density, specifically the high magnetic field strength provided by the permanent magnets, combined with the efficient control of two inverters, making them suitable for space-constrained vehicle environments. They also optimize energy regeneration, specifically the flexible control of energy flow by dual inverters. They also feature enhanced regenerative braking efficiency and redundant design. In the event of a single inverter failure, the other inverter can still drive the motor, enhancing vehicle safety.

[0003] An open-winding permanent magnet synchronous motor (PMSM) is equipped with two inverters, so its vector is the result of the combined output of the two inverters. The decoupling angle essentially adjusts the phase difference between the output voltage vectors of the dual inverters to achieve zero-sequence current suppression and efficient utilization of voltage resources. In a dual-inverter power supply system, improper coordination of the voltage vectors of the two inverters will lead to superposition of common-mode voltages, inducing zero-sequence circulating currents, increasing copper losses, reducing efficiency, and even threatening the safety of power devices. By introducing a dynamic decoupling angle, the voltage synthesis path can be optimized on the space vector plane, effectively offsetting zero-sequence voltage components while expanding the system's linear modulation range and improving DC bus voltage utilization. For example, traditional vector decoupling angle control strategies typically use a fixed 180° decoupling angle to maximize the number of output levels, but require closed-loop control to dynamically suppress zero-sequence current. In contrast, an adaptive decoupling angle strategy adjusts the phase difference in real time based on the load, balancing efficiency and dynamic response. The reasonable design of the decoupling angle not only affects the steady-state performance of the motor (such as torque ripple, total motor loss and current harmonic distortion rate), but also directly determines the fault-tolerant operation capability of the system under fault conditions.

[0004] Although the theoretical value of the decoupling angle has been widely recognized, its practical application still faces many challenges. First, the complex coupling relationship between the decoupling angle, modulation strategy, and zero-sequence current closed-loop control makes it difficult to establish an accurate mathematical model for multi-objective optimization. Second, different application scenarios have different requirements for the decoupling angle. For example, electric vehicle drive systems require fast dynamic response, while wind power grid-connected systems focus more on low harmonics and high stability, which requires a highly adaptable decoupling angle strategy. Current research focuses on the static analysis of a fixed decoupling angle, while exploration of adaptive adjustment of the dynamic decoupling angle and the integration of intelligent algorithms remains insufficient.

[0005] In response to the above problems, this paper focuses on the optimization design and control strategy of the vector decoupling angle of the open-winding permanent magnet synchronous motor. A dynamic decoupling angle cooperative modulation strategy is proposed to achieve high-performance output with reduced motor total loss, total harmonic distortion and torque pulsation. Summary of the Invention

[0006] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor.

[0007] To achieve the above object, the technical solution adopted by the present invention is:

[0008] A method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor comprises the following steps:

[0009] Step 1: Construct a mathematical model of an open-winding permanent magnet synchronous motor. Based on this mathematical model and the control target performance requirements, design a cost function for the vector decoupling angle of the open-winding permanent magnet synchronous motor. This cost function includes the motor output performance that the user needs to consider, such as torque error, total motor loss, and total harmonic current distortion.

[0010] Step 2: Set the application drive scenario of the open-winding permanent magnet synchronous motor and determine the value ranges of torque, speed, and vector decoupling angle;

[0011] Step 3: Based on the determined ranges of torque, speed, and vector decoupling angle, the different torques and decoupling angles at different speeds are traversed one by one, the operating points are scanned, and the torque error, total motor loss, and total harmonic current distortion rate corresponding to different vector decoupling angles under different operating conditions are calculated;

[0012] Step 4: Based on the calculated torque error, total motor loss, and total harmonic current distortion rate, and combined with the multi-objective Osprey optimization algorithm, a trained neural network model is obtained;

[0013] Step 5: The speed, torque, and vector decoupling angle under different working conditions are used as a second data set, and the second data set is input into the trained neural network model to obtain the predicted performance index data;

[0014] Step 6: Substitute the predicted performance index data under different working conditions into the cost function, calculate multiple cost function values, and then obtain the minimum cost function value and corresponding vector decoupling angle data under different working conditions, and build a database;

[0015] Step 7: Based on the constructed database, the dynamic optimization of the vector decoupling angle of the open-winding permanent magnet synchronous motor under different working conditions is achieved through the table lookup method.

[0016] Preferably, in step 1, the expression of the mathematical model of the open-winding permanent magnet synchronous motor is:

[0017]

[0018] Where i d with i q Denote the stator current of d-axis and q-axis respectively, u d with u q are the stator voltages of the d-axis and q-axis, respectively, and L d With L q They represent the equivalent inductance of the d-axis and q-axis respectively, R is the stator resistance, ω e is the electrical angular velocity, ψ f is the permanent magnet flux, and L is the filter inductance.

[0019] Preferably, in step 1, the cost function is expressed as:

[0020] J=10×|T e * -T e |+0.1×P loss +100×THD i

[0021] Where, T e * is the actual load torque, T e is the actual measured torque; P loss is the total motor loss, THD i is the total harmonic distortion of current.

[0022] Preferably, in step 2, the torque range is 0-60N·m, and the speed range is 0-1500r·min -1 , the value range of the vector decoupling angle is 60-300°.

[0023] Preferably, step 4 comprises the following steps:

[0024] Step 41: taking the calculated torque error, total motor loss, and total harmonic current distortion rate as a first data set, normalizing the first data set, and dividing the first data set into a training set and a validation set;

[0025] Step 42: Optimizing the weights and bias coefficients of the feedforward neural network based on the multi-objective Osprey optimization algorithm to obtain an optimized feedforward neural network;

[0026] Step 43: The optimized feedforward neural network is trained using the training set and verified using the validation set to obtain a trained neural network model.

[0027] Preferably, in step 42, the optimized feedforward neural network comprises an input layer, a hidden layer 1, a hidden layer 2, and an output layer connected in sequence;

[0028] The Sigmoid activation function is used to connect hidden layer 1 and hidden layer 2, and the Relu activation function is used to connect hidden layer 2 and the output layer.

[0029] Preferably, the input layer has 3 neurons, the hidden layer 1 has 11 neurons, the hidden layer 2 has 3 neurons, and the output layer has 3 neurons.

[0030] Preferably, in step 5, the predicted performance index data includes torque error, total motor loss and total harmonic distortion of current.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] (1) The present invention adopts the design of feedforward neural network based on multi-objective Osprey optimization algorithm to optimize the feedforward neural network. Compared with the traditional feedforward neural network, the neural network converges faster, has higher prediction accuracy, and improves the robustness and generalization of the neural network.

[0033] (2) The present invention can adjust the optimal vector decoupling angle in real time according to actual working conditions to improve the system output performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a flow chart of a method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor proposed in an embodiment of the present invention;

[0035] Figure 2 This is a design concept diagram of the predictive control weight factor of the open-winding permanent magnet synchronous motor mathematical model in an embodiment of the present invention;

[0036] Figure 3 1 is a driving system diagram of an open-winding permanent magnet synchronous motor in an embodiment of the present invention;

[0037] Figure 4 : is a loss model diagram of iron loss and copper loss of an open-winding permanent magnet synchronous motor in an embodiment of the present invention;

[0038] Figure 5 1 is a structural diagram of a feedforward neural network in an embodiment of the present invention;

[0039] Figure 6 is an evolutionary iteration graph of the multi-objective Osprey optimization algorithm in an embodiment of the present invention;

[0040] Figure 7 This is a comparison diagram of the effects of the feedforward neural network after multi-objective osprey optimization in an embodiment of the present invention and the traditional feedforward neural network;

[0041] Figure 8 This is a diagram of the optimized feedforward neural network training, prediction, and verification in an embodiment of the present invention;

[0042] Figure 9 A comparison diagram of the cost function J under working condition 1 provided by an embodiment of the present invention;

[0043] Figure 10 A comparison diagram of the cost function J under working condition 2 provided by an embodiment of the present invention;

[0044] Figure 11 The comparison diagram of the cost function J under working condition 3 is provided in the embodiment of the present invention;

[0045] Figure 12 A comparison diagram of the cost function J under working condition 4 provided by an embodiment of the present invention; DETAILED DESCRIPTION

[0046] The following is a diagram of the embodiment of the present invention. Figures 1 to 12 The technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0047] like Figure 1 and Figure 2 As shown, an embodiment of the present invention proposes a method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor, which specifically includes the following steps:

[0048] Step 1: Construct a mathematical model of an open-winding permanent magnet synchronous motor. Based on this mathematical model and the control target performance requirements, design a cost function for the vector decoupling angle of the open-winding permanent magnet synchronous motor. This cost function includes the motor output performance that the user needs to consider, such as torque error (Teerror), total motor loss (Ploss), and total harmonic current distortion (THDi).

[0049] Table 1 Parameters of open-winding permanent magnet synchronous motor

[0050]

[0051] like Figure 3 As shown, in the embodiment of the present invention, an open-winding permanent magnet synchronous motor is used as the control object, and its parameters are shown in Table 1. A mathematical model of the open-winding permanent magnet synchronous motor is constructed. The expression of the mathematical model of the open-winding permanent magnet synchronous motor (i.e., the stator voltage equation under the dq axis) is:

[0052]

[0053] Where i d with i q Represent the stator current of d-axis and q-axis respectively, u d with u q are the stator voltages of the d-axis and q-axis, respectively, and L d With L q They represent the equivalent inductance of the d-axis and q-axis respectively, and L is the filter inductance value. Since the equivalent inductance of the d-axis and q-axis of the surface-mount motor is equal, let L d =L q =L, R is the stator resistance, ω e is the electrical angular velocity, ψ f is the permanent magnet flux.

[0054] Then the torque expression in the dq coordinate system is:

[0055]

[0056] Among them, P n is the number of motor pole pairs, P dq Output power for the motor.

[0057] According to the established mathematical model of open-winding permanent magnet synchronous motor, the following is established: Figure 4 The loss model of motor iron loss and copper loss is shown in FIG. 1 . Copper loss is equivalent to the power consumed when the stator current flows through the stator resistance, and iron loss is equivalent to the power consumed when the iron loss branch current flows through the equivalent iron resistance. The loss model expressions of motor iron loss and copper loss are as follows:

[0058]

[0059] Among them, i wd and i wq are the useful work components of the d-axis and q-axis stator currents respectively.

[0060] When the system is stable, the speed and torque are constant, and the total motor loss can be expressed as:

[0061]

[0062] Among them, P loss is the total consumption of the motor, P Cu is the motor copper loss, P Fe is the motor iron loss, R c is the equivalent iron resistance, i cd and i cq are the iron loss components of the d-axis and q-axis stator currents respectively.

[0063] Derivation of i by Kirchhoff's current law and Kirchhoff's voltage law wd 、i wq 、i cd 、i cq The relationship between , the expression of total motor loss can be converted into:

[0064]

[0065] Among them, P loss is the total consumption of the motor, T e is the torque in the dq coordinate system.

[0066] The calculation method for the total harmonic distortion of current is:

[0067]

[0068] Where F1 is the fundamental amplitude and F is the total harmonic amplitude.

[0069] A cost function is defined to evaluate the performance of the selected vector decoupling angle. In order to optimize the overall performance of the motor, the torque error, total motor loss and total harmonic distortion are included in the cost function. Therefore, the cost function is expressed as follows:

[0070] J=10×|T e * -T e |+0.1×P loss +100×THD i

[0071] Where, T e * is the actual load torque, T e is the actual measured torque; the difference is Teerror , T eerror is the torque error, P loss is the total motor loss, THD i is the total harmonic distortion rate of current. The coefficients before the torque error, total motor loss and total harmonic distortion rate of current are to unify the three items in the expression into the range of 0-10 so as to be able to integrate the various output performances of the motor.

[0072] Step 2: Import the above model and motor parameters into Matlab, as shown below Figure 4 The simulation model is built as shown, and the application driving scenario of the open-winding permanent magnet synchronous motor is set to determine the value range of torque, speed and vector decoupling angle; the torque value range in this embodiment is 0-60N·m with a step size of 0.1875; the speed value range is 0-1500r·min -1 , with a step size of 10; the value range of the vector decoupling angle is 60-300°, with a step size of 1.

[0073] Step 3: Based on the determined ranges of torque, speed, and vector decoupling angle, the different torques and decoupling angles at different speeds are traversed one by one, the operating points are scanned, and the torque error, total motor loss, and total harmonic current distortion rate corresponding to different vector decoupling angles under different operating conditions are calculated;

[0074] Step 4: Based on the calculated torque error, total motor loss, and total harmonic current distortion rate, and combined with the multi-objective Osprey optimization algorithm, a trained neural network model is obtained. The specific steps include:

[0075] Step 41: The calculated torque error, total motor loss, and total harmonic current distortion rate are used as the first data set. After normalizing the first data set, the first data set is divided into a training set and a validation set; specifically, the first data set is divided into a training set and a validation set in a ratio of 7:3.

[0076] Step 42: Optimize the weights and bias coefficients of the feedforward neural network based on the multi-objective Osprey optimization algorithm to enable rapid convergence and reduce prediction error. Use the speed, torque, and decoupling angle as the input layer, and the torque error Teerror, the total motor loss Ploss, and the total harmonic current distortion rate THDi as the output layer. Minimize the root mean square error as the evaluation criterion. Finally, the combination of the motor operating condition and the decoupling angle is taken as a set of optimal values ​​to obtain the optimized feedforward neural network.

[0077] like Figure 5 and Figure 6As shown, the embodiment of the present invention applies a multi-objective osprey optimization algorithm to optimize the weights and bias coefficients of the feedforward neural network (randomly initialize the osprey population, global exploration (first stage: location identification and fishing) and local mining (second stage: bringing the fish to the appropriate location) to optimize the parameters of the feedforward neural network). Specifically, the population size is set to 100, the dimension is set to 3, and the boundary value is set to [-20, 20]. After the settings are completed, the multi-objective osprey optimization algorithm is used to optimize the weights and bias coefficients of the feedforward neural network. The number of training times is 1000, the learning rate is 0.01, and the minimum target is 0.00001. The optimized weights and bias coefficients are substituted into the feedforward neural network to obtain the optimized feedforward neural network. The minimum root mean square error RMSE is used as the evaluation index of the feedforward neural network.

[0078] In the embodiment of the present invention, the optimized feedforward neural network includes an input layer, a hidden layer 1, a hidden layer 2, and an output layer connected in sequence;

[0079] The Sigmoid activation function is used to connect hidden layer 1 and hidden layer 2, and the Relu activation function is used to connect hidden layer 2 and the output layer.

[0080] Specifically, the input layer has 3 neurons, the hidden layer 1 has 11 neurons, the hidden layer 2 has 3 neurons, and the output layer has 3 neurons.

[0081] In the embodiment of the present invention, the Sigmoid activation function is as follows:

[0082]

[0083] The Relu activation function is as follows:

[0084] f(x)=max(0,x)

[0085] The RMSE expression is as follows:

[0086]

[0087] Among them, y is the actual value, is the predicted value of the feedforward neural network.

[0088] Figure 7 and Figure 8 To optimize the process map, Figure 7 and Figure 8 It can be seen that the feedforward neural network optimized by the multi-objective Osprey optimization algorithm has a smaller error and is more stable. The feedforward neural network optimized by the multi-objective Osprey algorithm has an error within [-1, +1], which clearly shows the superiority of the present invention.

[0089] Step 43: Train the optimized feedforward neural network using the training set and verify it using the validation set. If the output network meets the requirements, that is, the error is less than 0.00001, the training can be stopped to obtain a trained neural network model. Otherwise, the learning rate, network structure and other parameters are adjusted and the training is continued until the output requirements are met.

[0090] Step 5: The speed, torque, and vector decoupling angle under different operating conditions are used as the second data set. This data set is input into the trained neural network model, and the output is the predicted performance indicator data, namely torque error, total motor loss, and total harmonic distortion rate of current.

[0091] Step 6: Substitute the predicted performance index data under different working conditions into the cost function, calculate multiple cost function values, and then obtain the minimum cost function value and corresponding vector decoupling angle data under different working conditions, and build a database;

[0092] Step 7: Dynamically optimize the open-winding permanent magnet synchronous motor's vector decoupling angle under different operating conditions using a table lookup method. Based on the constructed database, the vector decoupling angle corresponding to the minimum cost function value is retrieved based on the current operating conditions (i.e., different speeds and torques). This vector decoupling angle is then set as the vector decoupling angle for the open-winding permanent magnet synchronous motor, completing the dynamic optimization.

[0093] The present invention predicts the optimal vector decoupling angle for four different working conditions through the constructed database, and the results are shown in Table 2 below.

[0094] Table 2 Prediction results of optimal vector decoupling angle for 4 different working conditions

[0095]

[0096] Figures 9 to 12 The cost function J comparison chart under different working conditions is shown in Table 2 and Figures 9 to 12 The results show that the present invention verifies that the decoupling angle should be varied under different operating conditions and should not be fixed. Different vector decoupling angles should be designed under different operating conditions to achieve optimal performance of the open-winding permanent magnet synchronous motor drive system. This method is better than the traditional fixed vector decoupling angle of 180° and can adaptively adjust the decoupling angle after changes in operating conditions to maintain optimal system performance. Figures 9 to 12It can be seen that under working condition 1, the system performance of the present invention is improved by about 15% compared with the traditional decoupling angle control method; under working condition 2, the system performance of the present invention is improved by about 13% compared with the traditional decoupling angle control method; under working condition 3, the system performance of the present invention is improved by about 16% compared with the traditional decoupling angle control method; under working condition 4, the system performance of the present invention is improved by about 8% compared with the traditional decoupling angle control method. In summary, it can be seen that the effectiveness of the vector decoupling angle design method for the open-winding permanent magnet synchronous motor proposed in the present invention is achieved.

[0097] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the claims and their equivalents.

Claims

1. A method for designing the vector decoupling angle of an open-winding permanent magnet synchronous motor, characterized in that: The following steps are involved: Step 1: Construct a mathematical model of an open-winding permanent magnet synchronous motor and design a cost function for the vector decoupling angle of the open-winding permanent magnet synchronous motor based on the mathematical model of the open-winding permanent magnet synchronous motor; Step 2: Set the application drive scenario of the open-winding permanent magnet synchronous motor and determine the value ranges of torque, speed, and vector decoupling angle; Step 3: Based on the determined ranges of torque, speed, and vector decoupling angle, the different torques and decoupling angles at different speeds are traversed one by one, the operating points are scanned, and the torque error, total motor loss, and total harmonic current distortion rate corresponding to different vector decoupling angles under different operating conditions are calculated; Step 4: Based on the calculated torque error, total motor loss, and total harmonic current distortion rate, and combined with the multi-objective Osprey optimization algorithm, a trained neural network model is obtained; Step 5: The speed, torque, and vector decoupling angle under different working conditions are used as a second data set, and the second data set is input into the trained neural network model to obtain the predicted performance index data; Step 6: Substitute the predicted performance index data under different working conditions into the cost function, calculate multiple cost function values, and then obtain the minimum cost function value and corresponding vector decoupling angle data under different working conditions, and build a database; Step 7: Based on the constructed database, the dynamic optimization of the vector decoupling angle of the open-winding permanent magnet synchronous motor under different working conditions is achieved through the table lookup method.

2. The method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor according to claim 1, characterized in that: In step 1, the mathematical model of the open-winding permanent magnet synchronous motor is expressed as: Where i d with i q Denote the stator current of d-axis and q-axis respectively, u d with u q are the stator voltages of the d-axis and q-axis, respectively, and L d With L q They represent the equivalent inductance of the d-axis and q-axis respectively, R is the stator resistance, ω e is the electrical angular velocity, ψ f is the permanent magnet flux, and L is the filter inductance.

3. The method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor according to claim 1, characterized in that: In step 1, the cost function is expressed as: Where, is the actual load torque, T e is the actual measured torque; P loss is the total motor loss, THD i is the total harmonic distortion of current.

4. The method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor according to claim 1, wherein: In step 2, the torque range is 0-60N·m, and the speed range is 0-1500r·min -1 , the value range of the vector decoupling angle is 60-300°.

5. The method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor according to claim 1, characterized in that: Step 4 includes the following steps: Step 41: taking the calculated torque error, total motor loss, and total harmonic current distortion rate as a first data set, normalizing the first data set, and dividing the first data set into a training set and a validation set; Step 42: Optimizing the weights and bias coefficients of the feedforward neural network based on the multi-objective Osprey optimization algorithm to obtain an optimized feedforward neural network; Step 43: The optimized feedforward neural network is trained using the training set and verified using the validation set to obtain a trained neural network model.

6. The method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor according to claim 5, characterized in that: In step 42, the optimized feedforward neural network includes an input layer, a hidden layer 1, a hidden layer 2, and an output layer connected in sequence; The Sigmoid activation function is used to connect hidden layer 1 and hidden layer 2, and the Relu activation function is used to connect hidden layer 2 and the output layer.

7. The method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor according to claim 6, characterized in that: The input layer has 3 neurons, the hidden layer 1 has 11 neurons, the hidden layer 2 has 3 neurons, and the output layer has 3 neurons.

8. The method for designing a vector decoupling angle of an open-winding permanent magnet synchronous motor according to claim 1, characterized in that: In step 5, the predicted performance index data includes torque error, total motor loss and total harmonic distortion of current.

Images