A method and apparatus for suppressing harmonic currents in a motor

Abstract

The present application relates to a kind of motor harmonic current suppression method and device, belong to motor control technical field, wherein, the method includes: the difference of the direct axis current of motor and the average of direct axis current, obtain direct axis harmonic current, the difference of the cross axis current of motor and the average of cross axis current, obtain cross axis harmonic current;Direct axis harmonic current and cross axis harmonic current are input first PI controller, obtain first voltage compensation, direct axis harmonic current and cross axis harmonic current are input fuzzy controller, obtain second voltage compensation;First voltage compensation and second voltage compensation are weighted summation, obtain third voltage compensation, based on third voltage compensation, the output voltage of motor control system is compensated.The present application combines PI controller and fuzzy controller, with fuzzy controller supplementing the deficiency of PI control, improves the dynamic performance of control system.

Description

Technical Field

[0001] This invention relates to the field of motor control technology, and in particular to a method and apparatus for suppressing motor harmonic currents. Background Technology

[0002] After the introduction of an electric motor into the drive system of an electric vehicle, the original vibration characteristics are mixed with complex electromagnetic characteristics, resulting in a richer dynamic response of the drive system. Motor torque pulsation caused by current harmonics is the main factor leading to torsional vibration. The two main categories of causes of motor torque pulsation are not of the same nature, and numerous scholars both domestically and internationally have studied optimization schemes for each. For the motor body structure, the main optimization measures include using skewed slots, increasing rotor skew stages, replacing permanent magnets with different shapes, and changing the stator winding type. These measures help reduce torque pulsation to some extent, but they lack versatility, have high R&D costs, and cannot be optimized for motors already in use. It is precisely because of the inconvenience of optimizing the motor body structure that more and more researchers are focusing on optimizing motor control strategies.

[0003] For system control strategies, existing current harmonic suppression methods typically employ PI control strategies. The core of harmonic voltage compensation PI control strategies is to convert AC quantities into DC quantities to extract harmonic currents. However, multi-rotating coordinate system methods and complex vector methods often use low-pass filters to extract harmonic currents. Due to the influence of the cutoff frequency, high-precision low-pass filters exhibit strong delays, thus reducing the system's dynamic performance. Summary of the Invention

[0004] In view of this, it is necessary to provide a method and apparatus for suppressing motor harmonic currents to solve the technical problem of poor dynamic performance in the prior art.

[0005] To address the aforementioned problems, this invention provides a method for suppressing harmonic currents in a motor, comprising:

[0006] Obtain the three-phase current of the motor, perform coordinate transformation on the three-phase current to obtain the direct-axis current and quadrature-axis current;

[0007] The direct-axis harmonic current is obtained by subtracting the average value of the direct-axis current from the direct-axis current of the motor; the quadrature-axis harmonic current is obtained by subtracting the average value of the quadrature-axis current from the quadrature-axis current of the motor.

[0008] The direct-axis harmonic current and the quadrature-axis harmonic current are input into the first PI controller to obtain the first voltage compensation amount, and the direct-axis harmonic current and the quadrature-axis harmonic current are input into the fuzzy controller to obtain the second voltage compensation amount.

[0009] The first voltage compensation amount and the second voltage compensation amount are weighted and summed to obtain the third voltage compensation amount, and the output voltage of the motor control system is compensated based on the third voltage compensation amount.

[0010] In one possible implementation, the direct-axis current and the quadrature-axis current of the motor are obtained through the following steps:

[0011] Based on Clark transformation, the three-phase current of the motor in the ABC natural coordinate system is converted into the current in the two-phase stationary coordinate system. Based on Park transformation, the current in the two-phase stationary coordinate system is converted into the two-phase rotating coordinate system to obtain the direct-axis current and quadrature-axis current.

[0012] In one possible implementation, the fuzzy controller uses a triangular membership function.

[0013] In one possible implementation, the method further includes: optimizing the weight coefficients of the weighted summation of the PI controller and the fuzzy controller, as well as the parameters of the fuzzy controller, based on a reinforcement learning algorithm, with the optimization objectives of minimizing the harmonic current deviation, the rate of change of the harmonic current deviation, the amplitude of the third voltage compensation, and the total harmonic distortion.

[0014] In one possible implementation, the parameters of the fuzzy controller include the rule weights of the fuzzy controller and the membership function parameters of the fuzzy controller.

[0015] In one possible implementation, the reinforcement learning algorithm is a deep deterministic policy gradient algorithm.

[0016] In one possible implementation, the method further includes: inputting the deviation values ​​of the direct-axis harmonic current and quadrature-axis harmonic current from the desired current into a quasi-proportional multi-resonant controller to obtain a fourth voltage compensation amount, and compensating the output voltage of the motor control system based on the fourth voltage compensation amount and the third voltage compensation amount.

[0017] In one possible implementation, the quasi-proportional multi-resonant controller is composed of two quasi-proportional resonant controllers connected in parallel, wherein the resonant frequency of one quasi-proportional resonant controller is 6 times the fundamental frequency, and the resonant frequency of the other quasi-proportional resonant controller is 12 times the fundamental frequency.

[0018] In one possible implementation, the transfer function of the quasi-proportional resonant controller is:

[0019]

[0020] In the formula, Indicates proportional gain. Indicates the resonant gain. Indicates the resonant frequency. This indicates the cutoff frequency of the low-pass filter.

[0021] On the other hand, the present invention also provides a motor harmonic current suppression device, comprising:

[0022] The harmonic current extraction unit is used to obtain the direct-axis harmonic current by subtracting the direct-axis current of the motor from the average value of the direct-axis current, and to obtain the quadrature-axis harmonic current by subtracting the quadrature-axis current of the motor from the average value of the quadrature-axis current.

[0023] The compensation voltage injection unit is used to input the direct-axis harmonic current and the quadrature-axis harmonic current into the first PI controller to obtain a first voltage compensation amount, input the direct-axis harmonic current and the quadrature-axis harmonic current into the fuzzy controller to obtain a second voltage compensation amount, perform a weighted summation of the first voltage compensation amount and the second voltage compensation amount to obtain a third voltage compensation amount, and compensate the output voltage of the motor control system based on the third voltage compensation amount.

[0024] The beneficial effects of this invention are as follows: The motor harmonic current suppression method provided by this invention first uses the current averaging method to extract the full-order harmonic current, avoiding the problem of the dynamic performance being affected by the use of a low-pass filter to extract the harmonic current. Then, it utilizes the fuzzy controller, which can adjust the control strategy according to the real-time state of the system and has good adaptability. By combining the PI controller and the fuzzy controller to obtain the compensation voltage, the fuzzy controller can supplement the shortcomings of the PI control when the system changes or there are uncertainties, effectively improving the dynamic performance of the control system. Attached Figure Description

[0025] Figure 1 A schematic flowchart of an embodiment of the motor harmonic current suppression method provided by the present invention;

[0026] Figure 2 A schematic diagram of the motor control system provided by the present invention;

[0027] Figure 3 This is a schematic diagram of the reinforcement learning logic flow provided by the present invention;

[0028] Figure 4 This is a schematic diagram of the training steps provided by the present invention;

[0029] Figure 5 The schematic diagram of the QPMR controller structure provided by this invention;

[0030] Figure 6 This is a schematic diagram of an embodiment of the motor harmonic current suppression device provided by the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0032] It should be understood that the illustrative drawings are not drawn to scale. The flowcharts used in this invention illustrate operations implemented according to some embodiments of the invention. It should be understood that the operations in the flowcharts may be implemented out of order, and steps without logical contextual relationships may be reversed or performed simultaneously. Furthermore, those skilled in the art, guided by the content of this invention, may add one or more other operations to the flowcharts, or remove one or more operations from the flowcharts. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor systems and / or microcontroller systems.

[0033] The terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. "And / or" describes the relationship between related objects, indicating that three relationships may exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone.

[0034] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0035] Before demonstrating the embodiments, the following terms will be explained.

[0036] This invention provides a method and apparatus for suppressing harmonic currents in a motor, which will be described below.

[0037] Figure 1 A schematic flowchart of an embodiment of the motor harmonic current suppression method provided by the present invention is shown below. Figure 1 As shown, the methods for suppressing motor harmonic currents include:

[0038] S101. Subtract the direct-axis current from the average value of the direct-axis current to obtain the direct-axis harmonic current; subtract the quadrature-axis current from the average value of the quadrature-axis current to obtain the quadrature-axis harmonic current.

[0039] To better decouple the three-phase currents and facilitate subsequent control, in some embodiments of the present invention, the direct-axis current and quadrature-axis current of the motor are obtained through the following steps: first, the three-phase currents in the ABC natural coordinate system are converted using Clark transformation. and The current in a two-phase stationary coordinate system is used to simplify the three-phase system into a two-phase system, reducing the number of variables and simplifying the model. Then, the Park transformation is applied to... and Current conversion in two-phase stationary coordinate system to and In a two-phase rotating coordinate system, alternating currents are converted into direct currents, resulting in direct-axis and quadrature-axis currents, thus achieving complete decoupling of the three-phase currents. It is important to note that... Indicates the direct shaft of the motor. This indicates the quadrature axis of the motor.

[0040] It should be noted that the explanation and calculation are based on a permanent magnet synchronous motor. The mathematical model of a permanent magnet synchronous motor in a two-phase rotating coordinate system includes voltage equations, flux linkage equations, and electromagnetic torque equations. The voltage equation is as follows:

[0041]

[0042] In the formula, This represents the direct-axis voltage component of the motor. This represents the quadrature-axis voltage component of the motor. This represents the direct-axis current component of the motor. This represents the quadrature-axis current component of the motor. This represents the direct-axis inductance of the motor. This represents the quadrature-axis inductance of the motor. This represents the phase resistance of the motor stator. This indicates the electric angular velocity of the motor.

[0043] The flux linkage equation is:

[0044]

[0045] In the formula, This represents the direct-axis flux linkage component of the motor. This represents the cross-axis flux linkage component of the motor. This indicates the magnetic flux linkage of a permanent magnet.

[0046] The electromagnetic torque equation is:

[0047]

[0048] In the formula, This represents the electromagnetic torque of the motor. This indicates the number of pole pairs of the motor.

[0049] It should also be noted that the formula for extracting harmonic current is:

[0050]

[0051]

[0052] In the formula, Represents direct-axis current. Indicates quadrature-axis current. This represents the average value of the fundamental frequency of the direct-axis current. This represents the average value of the fundamental frequency of the quadrature-axis current. Indicates any time. Indicates the fundamental frequency period. This represents the sum of direct-axis currents, i.e., direct-axis harmonic currents. This represents the sum of quadrature-axis currents, i.e., quadrature-axis harmonic currents.

[0053] S102. Input the direct-axis harmonic current and the quadrature-axis harmonic current into the first PI controller to obtain the first voltage compensation amount, and input the direct-axis harmonic current and the quadrature-axis harmonic current into the fuzzy controller to obtain the second voltage compensation amount.

[0054] It should be noted that the fuzzy controller uses a triangular membership function, the expression of which is:

[0055]

[0056] In the formula, This represents the left base point of the triangle. Indicates the center point of the triangle. Let represent the right base point of the triangle. Then, for multiple fuzzy sets, the parameter set is: .

[0057] Fuzzy control statements are based on " "Formulation, in the formula, This indicates the amount of harmonic current deviation. The value represents the rate of change of harmonic current deviation; the input to the fuzzy controller is defined as d, which is the deviation between the target value and the value of the quadrature-axis harmonic current. Harmonic current deviation rate of change The output is the harmonic voltage compensation amount. ;set up The basic domain of discourse is , The basic domain of discourse is , The domain of discourse is ;set up The domain of the fuzzy subset is , The fuzzy subset domain is , The fuzzy subset domain is The quantization factor and scaling factor of the input quantity are expressed as follows:

[0058]

[0059] In the formula, and Indicates the quantification factor. This represents the scaling factor.

[0060] Fuzzy control statements are based on " "Formulate, assuming the system has" There are 10 rules, and each rule is assigned a weight. Then the output of the fuzzy controller becomes:

[0061]

[0062] In the formula, Representation rules The activation function, Representation rules The weight.

[0063] S103. The first voltage compensation amount and the second voltage compensation amount are weighted and summed to obtain the third voltage compensation amount. The output voltage of the motor control system is compensated based on the third voltage compensation amount.

[0064] It should be noted that the formula for calculating the third voltage compensation is as follows:

[0065]

[0066] In the formula, This indicates the third voltage compensation amount. This represents the first voltage compensation amount output by the PI controller. This represents the second voltage compensation value output by the fuzzy controller. and This represents the adaptive weighting coefficients, satisfying the following conditions: .

[0067] Based on the aforementioned fuzzy controller, the first PI controller, and the corresponding algorithm, a fuzzy-PI composite controller is constructed. The inputs of the fuzzy-PI composite controller are the direct-axis current and the quadrature-axis current. , The output is the third voltage compensation amount ( It should be noted that the direct-axis current and quadrature-axis current are processed separately, and the third voltage compensation amount... Including the third voltage compensation amount on the direct axis Voltage compensation amount of the cross axis ; in such Figure 2 In the motor control system shown, the desired speed is ( The current is converted into the desired direct-axis current and quadrature-axis current via a second PI controller and MTPA-FWC. , ), expected direct-axis current and quadrature-axis current ( , ) and actual direct-axis current and quadrature-axis current ( , The deviation value is input to the third PI controller, which is the basic controller for the motor. The third PI controller outputs the basic voltage, and simultaneously the direct-axis current and quadrature-axis current. , It also inputs to the fuzzy-PI composite controller, and the fuzzy-PI composite controller outputs a third voltage compensation quantity. ), through the third voltage compensation amount ( The base voltage is compensated to determine the final output voltage for motor control.

[0068] To better tune the weight coefficients of the PI controller and the fuzzy controller, as well as the parameters of the fuzzy controller, in some embodiments of the present invention, the method further includes: based on reinforcement learning, using the harmonic current deviation... Harmonic current deviation rate of change Third voltage compensation amplitude The optimization objective is to minimize the total harmonic distortion (THD). The weighting coefficients of the PI controller and the fuzzy controller (i.e.,...) are... , ), and the parameters of the fuzzy controller (including the rule weights of the fuzzy controller). Membership function parameters of the fuzzy controller ) is optimized, and the optimized function expression is:

[0069]

[0070] In the formula, , , and This represents the weighting coefficient.

[0071] It should be noted that the logical flow of reinforcement learning optimization methods is as follows: Figure 3As shown, it mainly consists of a state space (State), an action space (Action), and a reward function (Reward). The State includes: harmonic current deviation, harmonic current deviation rate of change, third voltage compensation, and total harmonic distortion. The Action includes: adjusting the output ratio combination of the PI controller and the fuzzy controller, i.e. , Modify the membership function parameters of the fuzzy controller. and rule weight The main considerations for setting the reward function are: control. To ensure accuracy; control To avoid excessive system oscillation; control Control system energy consumption and power loss; control The goal is to minimize harmonic distortion.

[0072] Reward can be represented as:

[0073]

[0074] The Deep Reinforcement Learning (DDPG) algorithm, also known as the Deep Deterministic Policy Gradient (DDPG) algorithm, was chosen as the method for training the agent. It combines the Actor-Critic method (Deterministic Policy Gradient, DPG) with a deep neural network, using two networks (a policy network and a value network) and a target network for stable training.

[0075] The initialization phase includes initializing network parameters, initializing the experience replacement pool, defining the environment, and setting hyperparameters; among these, initializing network parameters includes initializing the Actor network. and Critic Network The weights, and the parameters of the 2Actor and Critic networks, are then assigned to the corresponding target network: , Initialize the experience replacement pool: Set the experience replacement pool size to M; Define the environment: Create a Gym and obtain its initial state. ; Set hyperparameters: Actor network learning rate Critic network learning rate Discount factor Soft update coefficient .

[0076] The steps of the training phase are as follows Figure 4 As shown, the action selection includes: selecting an action based on the current state and the Actor network, and adding noise for exploration, expressed as: In the formula, This represents exploratory noise, typically OU (Ornstein-Uhlenbeck) noise; actions performed include: performing actions within the environment. Get the next state ,award and termination mark ;Storing experience includes: storing experience Stored in the experience replacement pool; mini-batch sampling includes: randomly sampling mini-batches from the experience replacement pool. Updating the Critic network includes: calculating the target Q-value. And minimizing the Critic network loss: Updating the Actor network includes updating the Actor network parameters via gradient ascent. Soft updates to the target network include updating the parameters of the Actor and Critic target networks. The termination conditions for the check include: setting a termination flag. Total Harmonic Distortion (THD) is the value if the current round ends. Training ends when the number of training rounds falls below a certain threshold or reaches the maximum number of training rounds.

[0077] In summary, the motor harmonic current suppression method of the present invention first uses the current averaging method to extract the full-order harmonic current, avoiding the problem of the impact on dynamic performance caused by using a low-pass filter to extract harmonic current. Then, it utilizes the fuzzy controller, which can adjust the control strategy according to the real-time state of the system and has good adaptability. By combining the PI controller and the fuzzy controller to obtain the compensation voltage, the fuzzy controller can supplement the shortcomings of the PI control when the system changes or there are uncertainties, effectively improving the dynamic performance of the control system.

[0078] Considering that the main sources of harmonic current are harmonic current caused by inverter nonlinearity and harmonic current caused by bias current, the harmonic current caused by inverter nonlinearity is:

[0079]

[0080] In the formula, This represents the direct-axis deviation current component. This represents the cross-axis deviation current component. Indicates the deviation voltage. Indicates the harmonic order. Indicates the stator phase resistance. Represents electric angular velocity. Indicates the current-controlled angle.

[0081] The harmonic current caused by the bias current is:

[0082]

[0083] In the formula, , These represent the direct and quadrature axis deviation current components, respectively. , This represents the bias current of phase A and phase B. , This indicates the sampling error of the current in phase A and phase B.

[0084] The harmonic current orders are thus summarized in Table 1.

[0085] Table 1: Sources and Order Distribution of Harmonic Currents in Motors

[0086]

[0087] Based on the analysis in Table 1, this invention combines a first PI controller with a fuzzy controller to eliminate full-order harmonic currents, but it is not suitable for a large number of 6 k For example, the elimination effect of harmonic currents of the 6th and 12th orders is not good. Therefore, in order to suppress specific order harmonic currents and improve the harmonic current suppression effect, in the embodiments of the present invention, the method further includes: inputting the deviation values ​​of the direct-axis harmonic current and quadrature-axis harmonic current from the desired current into a quasi-proportional multi-resonant controller to obtain a fourth voltage compensation amount, and compensating the output voltage of the motor control system based on the fourth voltage compensation amount and the third voltage compensation amount.

[0088] It should be noted that the quasi-proportional multiresonant controller (QPMR) is composed of two quasi-proportional resonant (QPR) controllers connected in parallel. The quasi-proportional resonant controller is obtained by optimizing the PR controller by replacing the integrator with a first-order low-pass filter. Its transfer function is:

[0089]

[0090] because The transfer function can be simplified to:

[0091]

[0092] In the formula, Indicates proportional gain. Indicates the resonant gain. Indicates the resonant frequency. This indicates the cutoff frequency of the low-pass filter.

[0093] Connect the two QPR controllers in parallel to form a QPMR controller, and set them respectively. , , The fundamental frequency is represented by the sub-controller, which shares both integral and proportional elements to avoid multiple parameter tuning steps and simultaneously suppresses multiple harmonic currents. Its structural principle is as follows: Figure 5 As shown.

[0094] In such Figure 2 In the motor control system shown, the QPMR controller is connected in parallel with the third PI controller. It's important to note that the direct-axis current and quadrature-axis current are processed separately. Figure 2 There are two third PI controllers, one corresponding to the direct axis and the other to the quadrature axis. Correspondingly, each third PI controller is connected in parallel with another third PI controller. One third PI controller is used to output the fourth voltage compensation amount for the direct axis. Another third PI controller is used to output the fourth voltage compensation amount of the quadrature axis. The desired direct-axis current and quadrature-axis current ( , ) and actual direct-axis current and quadrature-axis current ( , The deviation value is input to the QPMR controller, and the QPMR controller outputs the fourth voltage compensation value ( , ), fourth voltage compensation amount ( ) and third voltage compensation amount ( , Together, they compensate for the voltage output by the third PI controller to obtain the final output voltage, which is then used to control the motor.

[0095] Considering that the next state is obtained by performing actions in the environment during the reinforcement learning training phase, it can be obtained by simulating the motor model. The main sources of harmonic current are harmonic current caused by inverter nonlinearity and harmonic current caused by bias current. Therefore, in some embodiments of the present invention, when constructing the motor model, the harmonic current caused by the nonlinearity of the motor control system is considered. The Transport Delay module is added to the six-phase PWM wave of the motor model to simulate the delay caused by the "dead zone effect". The conduction voltage drop and freewheeling voltage drop are set in the motor model to simulate the conduction voltage drop of electronic components. Current components are added to the A and C phases of the motor model to simulate the bias current.

[0096] To better implement the motor harmonic current suppression method in this embodiment of the invention, based on the existing motor harmonic current suppression method, correspondingly, as follows: Figure 6 As shown, this embodiment of the invention also provides a motor harmonic current suppression device 600, comprising:

[0097] The harmonic current extraction unit 602 is used to subtract the direct-axis current of the motor from the average value of the direct-axis current to obtain the direct-axis harmonic current, and to subtract the quadrature-axis current of the motor from the average value of the quadrature-axis current to obtain the quadrature-axis harmonic current.

[0098] The compensation voltage injection unit 603 is used to input the direct-axis harmonic current and the quadrature-axis harmonic current into the first PI controller to obtain the first voltage compensation amount, input the direct-axis harmonic current and the quadrature-axis harmonic current into the fuzzy controller to obtain the second voltage compensation amount, and perform a weighted summation of the first voltage compensation amount and the second voltage compensation amount to obtain the third voltage compensation amount. The output voltage of the motor control system is compensated based on the third voltage compensation amount.

[0099] The motor harmonic current suppression device 600 provided in the above embodiments can realize the technical solution described in the above embodiments of the motor harmonic current suppression method. The specific implementation principle of each unit can be found in the corresponding content in the above embodiments of the motor harmonic current suppression method, and will not be repeated here.

[0100] The present invention provides a detailed description of a motor harmonic current suppression method. Specific examples have been used to illustrate the principle and implementation of the invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​the invention. At the same time, those skilled in the art will know that there will be changes in the specific implementation and application scope based on the idea of ​​the invention. Therefore, the content of this specification should not be construed as a limitation of the invention.

[0101] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method of motor harmonic current suppression, characterized by, The method comprises the following steps: differences between the direct-axis current of the motor and an average value of the direct-axis current to obtain a direct-axis harmonic current, and differences between the cross-axis current of the motor and an average value of the cross-axis current to obtain a cross-axis harmonic current; the direct-axis harmonic current and the cross-axis harmonic current are full-order harmonic currents; the direct-axis harmonic current and the cross-axis harmonic current are input into a first PI controller to obtain a first voltage compensation amount, and the direct-axis harmonic current and the cross-axis harmonic current are input into a fuzzy controller to obtain a second voltage compensation amount; the first voltage compensation amount and the second voltage compensation amount are weighted and summed to obtain a third voltage compensation amount, and an output voltage of the motor control system is compensated based on the third voltage compensation amount; Wherein, based on the reinforcement learning algorithm, the harmonic current deviation , the harmonic current deviation rate , the third voltage compensation amount amplitude and the total harmonic distortion amount The minimum is the optimization target, the weight coefficient of the weighted sum of the PI controller and the fuzzy controller , and the parameters of the fuzzy controller, including the rule weight of the fuzzy controller and the membership function parameters of the fuzzy controller The optimization function expression is: In the formula, , , and represent weight coefficients, T is the fundamental period.

2. The method of harmonic current suppression for electrical machines according to claim 1, characterized in that, the direct-axis current of the motor and the cross-axis current of the motor are obtained through the following steps: based on a Clark transformation, three-phase currents of the motor in an ABC natural coordinate system are converted into currents in a two-phase static coordinate system, and based on a Park transformation, the currents in the two-phase static coordinate system are converted into direct-axis currents and cross-axis currents in a two-phase rotating coordinate system.

3. The method of harmonic current suppression in electric machines according to claim 1, characterized in that, The fuzzy controller selects a triangular membership function.

4. The method of harmonic current suppression for electric machines according to claim 1, characterized in that, The reinforcement learning algorithm is a deep deterministic policy gradient algorithm.

5. The method of harmonic current suppression in electric machines according to claim 1, characterized in that, The method further comprises the following steps: the direct-axis harmonic current and the cross-axis harmonic current and a deviation value of an expected current are input into a quasi-proportional resonant controller to obtain a fourth voltage compensation amount, and an output voltage of the motor control system is compensated based on the fourth voltage compensation amount and the third voltage compensation amount.

6. The method of harmonic current suppression for electric machines according to claim 5, characterized in that, The quasi-proportional resonant controller is composed of two quasi-proportional resonant controllers in parallel, one of which has a resonant frequency of 6 times a fundamental frequency, and the other has a resonant frequency of 12 times the fundamental frequency.

7. A method of harmonic current suppression for an electrical machine according to claim 6, characterized in that, The transfer function of the quasi-proportional resonant controller is: wherein represents a proportional gain, represents a resonant gain, represents a resonant frequency, represents a cut-off frequency of a low-pass filter.

8. An apparatus for suppressing harmonic currents in an electric machine, characterized by The method comprises the following steps: a harmonic current extraction unit is configured to obtain a direct-axis harmonic current by subtracting an average value of the direct-axis current of the motor from the direct-axis current, and obtain a cross-axis harmonic current by subtracting an average value of the cross-axis current of the motor from the cross-axis current; the direct-axis harmonic current and the cross-axis harmonic current are full-order harmonic currents; a compensation voltage injection unit is configured to input the direct-axis harmonic current and the cross-axis harmonic current into a first PI controller to obtain a first voltage compensation amount, input the direct-axis harmonic current and the cross-axis harmonic current into a fuzzy controller to obtain a second voltage compensation amount, and weight and sum the first voltage compensation amount and the second voltage compensation amount to obtain a third voltage compensation amount; and an output voltage of the motor control system is compensated based on the third voltage compensation amount. Wherein, based on the reinforcement learning algorithm, the harmonic current deviation , the harmonic current deviation rate , the third voltage compensation amount amplitude and the total harmonic distortion amount The minimum is the optimization target, the weight coefficient of the weighted sum of the PI controller and the fuzzy controller , and the parameters of the fuzzy controller, including the rule weight of the fuzzy controller and the membership function parameters of the fuzzy controller The optimization function expression is: wherein , , and denote weight coefficients, T is the fundamental period.

Images