A motor torque fluctuation suppression method based on CNN-LSTM

Abstract

The application discloses a motor torque fluctuation suppression method based on CNN-LSTM, comprising the following steps: S1, establishing a mathematical model of a permanent magnet synchronous motor; S2, establishing an inverter model considering inverter dead zones and tube voltage drops; S3, establishing a harmonic compensation voltage prediction model based on CNN-LSTM to optimize DQ-axis stator voltage, reduce current harmonics, suppress motor vibration and torque fluctuation; S4, embedding the harmonic compensation voltage prediction model based on CNN-LSTM of step S3 into a motor control system to suppress torque fluctuation of the motor. The application effectively optimizes DQ-axis stator voltage, reduces 5th and 7th current harmonics which have greater influence on torque fluctuation, suppresses torque fluctuation, reduces control parameters, improves system response speed, and enhances the suppression effect of the motor on torque fluctuation at high speed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of permanent magnet synchronous motor, and particularly relates to a motor torque fluctuation suppression method based on CNN-LSTM. BACKGROUND

[0002] Permanent magnet synchronous motor (PMSM) has been widely used in electric drive system of electric vehicles due to its high power density, high efficiency and wide speed regulation range. However, the torque fluctuation of permanent magnet synchronous motor in the electric drive system is an important reason for the problem of electromagnetic noise and torsional vibration. On the one hand, the electromagnetic noise of the electric drive system is different from the noise of the traditional gasoline engine, and the frequency is higher, which is easy to make people feel uncomfortable. On the other hand, serious torque fluctuation of the motor will cause the increase of hysteresis and eddy current loss of the motor, so that the electric stress of the motor material increases, and the service life of the motor is shortened. The smoothness of torque output becomes one of the important indicators to measure the dynamic performance and steady-state performance of motor control, and how to suppress the torque fluctuation of permanent magnet synchronous motor has become one of the current research hotspots.

[0003] In recent years, scholars have devoted to the research on the influence of inverter dead time on output current harmonics. From the control point of view, the following three methods are mainly used to suppress the current harmonic distortion and torque fluctuation caused by inverter dead time effect. The first method is iterative learning control method. The iterative learning control method is a model-free control method which "learns" the periodic deviation signal in a memory way. The second method is model compensation method. The voltage distortion model caused by the dead time of voltage source inverter (VSI) is established, and the feedforward control is used for compensation. The third method is current harmonic injection method. The control algorithm is used to suppress the distortion of current harmonics.

[0004] The above methods have their own limitations. The iterative learning control method is difficult to provide accurate compensation information when the speed is not constant. The model compensation method depends on the judgment of current direction, and the judgment error is easy to occur near the zero point of current. The current harmonic injection method increases the PI controller, and the design is difficult due to more adjustment parameters. Moreover, the above three methods have poor suppression effect on the torque fluctuation of the motor at high speed. SUMMARY

[0005] In order to solve the above technical problems, the present application provides a motor torque fluctuation suppression method based on CNN-LSTM. The harmonic compensation voltage prediction model based on CNN-LSTM is established to optimize the DQ axis stator voltage, reduce the current harmonics, suppress the torque fluctuation, reduce the control parameters, improve the system response speed, and enhance the suppression effect of the motor on the torque fluctuation at high speed.

[0006] The technical scheme adopted by the present application to solve its technical problems is: a motor torque fluctuation suppression method based on CNN-LSTM, comprising the following steps:

[0007] S1: establishing a mathematical model of the permanent magnet synchronous motor;

[0008] S2: establishing an inverter model considering the inverter dead zone and tube voltage drop;

[0009] S3: establishing a harmonic compensation voltage prediction model based on CNN-LSTM to optimize the DQ axis stator voltage, reduce current harmonics, and suppress motor vibration and torque fluctuation; specifically comprising the following steps:

[0010] S31: establishing a current harmonic model;

[0011] S32: collecting training data;

[0012] S33: establishing a harmonic compensation voltage prediction model based on CNN-LSTM;

[0013] S4: embedding the harmonic compensation voltage prediction model based on CNN-LSTM of step S3 into the motor control system to suppress the torque fluctuation of the motor.

[0014] Further, the step S31, the specific method of establishing a current harmonic model is as follows:

[0015] The three-phase voltage u a , u b , u c of the permanent magnet synchronous motor is transformed into the DQ rotating coordinate system to obtain:

[0016]

[0017] In the formula, u d1 and u q1 are the d-axis and q-axis components of the fundamental voltage in the DQ rotating coordinate system, u5 and u7 are the amplitudes of the 5th and 7th harmonic voltages in the DQ rotating coordinate system, θ5 and θ7 are the initial phase angles of the 5th and 7th harmonic voltages, and ω is the rotor angular velocity;

[0018] The mathematical equation of the motor current in the DQ rotating coordinate system is:

[0019]

[0020] In the formula, i d1 and i q1i d and i q are the d-axis and q-axis components of the fundamental current in the DQ rotating coordinate system, i 5 and i 7 are the amplitudes of the 5th and 7th harmonic currents in the DQ rotating coordinate system, θ 5 and θ 7 are the initial phase angles of the 5th and 7th harmonic voltages, and ω is the rotor electric angular velocity;

[0021] The motor stator voltage equation containing harmonic components is:

[0022]

[0023] In the formula, i d1 and i q1 are the d-axis and q-axis components of the fundamental current in the DQ rotating coordinate system, i 5 and i 7 are the amplitudes of the 5th and 7th harmonic currents in the DQ rotating coordinate system, R s is the stator resistance, L d and L q are the d-axis and q-axis inductances, θ 5 and θ 7 are the initial phase angles of the 5th and 7th harmonic voltages, and ω is the rotor electric angular velocity;

[0024] Since the fundamental is an alternating component in the 5th and 7th harmonic DQ rotating coordinate system, the harmonic steady-state voltage equation in the 5th DQ rotating coordinate system is:

[0025]

[0026] In the formula, i d5 and i q5 are the d-axis and q-axis direct current components in the 5th DQ rotating coordinate system, R s is the stator resistance, L d and L q are the d-axis and q-axis inductances, and ω is the rotor electric angular velocity;

[0027] Similarly, the harmonic steady-state voltage equation in the 7th DQ rotating coordinate system is:

[0028]

[0029] In the formula, i d7 and i q7 are the d-axis and q-axis direct current components in the 7th DQ rotating coordinate system, R s is the stator resistance, L d and L q are the d-axis and q-axis inductances, and ω is the rotor electric angular velocity;

[0030] The three-phase motor currents pass through the 5th and 7th rotating coordinate systems of the DQ axis, and the 5th and 7th current harmonics in the DQ axis are obtained by using low-pass filtering;

[0031] The harmonic compensation voltage in the DQ coordinate system is obtained by summing the harmonic voltages of each current respectively

[0032] The harmonic compensation voltage in the DQ coordinate system is obtained by summing the harmonic voltages of each current respectively And

[0033] That is:

[0034] Further, the step S32: the specific method for collecting training data is that the initial rotating speed of the motor is set as 50 rpm, the motor torque is gradually increased from 0Nm to the maximum torque, and the increment is set as 10Nm; the data with a time length of 1 second is collected at each working condition; the rotating speed is increased to the maximum speed with an increment of 50 rpm, and the data is repeatedly recorded at each working condition; the d-axis current i d , the q-axis current i q , the rotor electric angle speed θ, the rotor angle ω and the motor reference torque are recorded at each working condition; then a large amount of experimental data is normalized and used for training the CNN-LSTM.

[0035] Further, the step S33, the specific method for establishing the harmonic compensation voltage prediction model based on the CNN-LSTM is that:

[0036] The convolutional neural network and the long short-term neural network are combined to establish the CNN-LSTM network model; the training of the CNN-LSTM network includes two stages:

[0037] In the first stage, i d , i q , θ, ω and are input into the convolutional neural network; in the second stage, the features extracted by the convolutional neural network are transmitted to the long short-term neural network, and the full connection layer is used as the prediction result;

[0038] The convolutional neural network is used for initial feature recognition of the input data, then the long short-term neural network is used for iterative training of the screened feature values, and the final output prediction result is the harmonic compensation voltage; the mapping relationship with i d , i q , θ, ω as input variables and as output is established. And

[0039] The beneficial effects of the present application compared with the prior art are:

[0040] ​The application aims at the torque fluctuation problem of a vehicle permanent magnet synchronous motor caused by the non-linear factors of an inverter, and proposes a torque fluctuation suppression method based on CNN-LSTM; the method fully considers the current harmonics caused by the dead zone and tube voltage drop of the inverter, reduces the 5th and 7th current harmonics which have greater influence on torque fluctuation through a harmonic compensation voltage prediction model based on CNN-LSTM, suppresses the torque fluctuation, reduces the control parameters, improves the system response speed, and enhances the suppression effect of the motor on torque fluctuation at high speed. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a flow chart of a motor torque fluctuation suppression method based on CNN-LSTM of the application.

[0042] Figure 2 is a topology diagram of the inverter in the application.

[0043] Figure 3 is a voltage waveform diagram of the A-phase bridge arm in the application.

[0044] Figure 4 is a 5th and 7th current harmonic extraction module of the permanent magnet synchronous motor.

[0045] Figure 5 is a 5th and 7th current harmonic decoupling control block diagram of the permanent magnet synchronous motor.

[0046] Figure 6 is a CNN-LSTM network.

[0047] Figure 7 is a training process flow chart of the CNN-LSTM.

[0048] Figure 8 is a principle diagram of the motor torque fluctuation suppression method based on CNN-LSTM. DETAILED DESCRIPTION

[0049] The application will be further described in detail below with reference to the accompanying drawings. It should be understood that the preferred embodiments are only for illustrating the application, and are not intended to limit the protection scope of the application. Among them, the drawings are only used for illustrative description, and cannot be understood as limiting the application; it is understandable to those skilled in the art that some well-known structures and their descriptions in the drawings can be omitted.

[0050] The application discloses a motor torque fluctuation suppression method based on CNN-LSTM.

[0051] Reference Figure 1 The motor torque fluctuation suppression method based on CNN-LSTM includes the following steps:

[0052] S1: Establish a mathematical model of the permanent magnet synchronous motor.

[0053] In this embodiment, the basic parameters of the selected permanent magnet synchronous motor are: maximum power 160kW, maximum torque 320Nm, maximum speed 16000Rpm, DC voltage 354V, number of pole pairs 4, stator flux linkage 0.0255Wb, and stator inductance 0.05mH.

[0054] In a synchronous rotating coordinate system, the voltage equation of a permanent magnet synchronous motor is:

[0055]

[0056] In the formula, u d and u q These are the d-axis and q-axis voltages, i d and i q These are the d-axis and q-axis currents, respectively, ψ d and ψ q Let R be the components of the stator flux linkage along the d-axis and q-axis, respectively, and ω be the rotor electric angular velocity. s This is the stator resistance.

[0057] S2: Establish an inverter model that considers the inverter dead zone and tube voltage drop.

[0058] Inverter circuits used in electric vehicles, such as Figure 2 As shown. Define S. a S b S c S a ′、S b ′、S c '' represent the switching states of the inverter. The inverter works as follows: when S... a / S b / S c When S is in state 1 a ′ / S b ′ / S c When S is in the 0 state, the three switching devices on the upper bridge arm of the inverter circuit are turned on, and the three switching devices on the lower bridge arm of the inverter circuit are turned off. Conversely, when S is in the 0 state... a ′ / S b ′ / S c When ′ is in state 1, S a / S b / S c When the state is 0, the three switching devices of the upper bridge arm of the inverter circuit are turned off, and the three switching devices of the lower bridge arm of the inverter circuit are turned on.

[0059] Taking into account both the inverter dead time and the transistor voltage drop, the actual output voltage waveform of phase A is as follows: Figure 3 As shown in the figure. When the A-phase current i a When >0, the output voltage fluctuation of phase A bridge arm is:Figure 3 (a). A-phase current i a When <0, the output voltage fluctuation of A-phase bridge arm is Figure 3 (b). In the figure, v t is the on voltage drop of the switch tube, v d is the on voltage drop of the freewheeling diode, T d is the dead time.

[0060] Assume that in a pulse width modulation cycle, set T S1 , T S2 are the actual on time of the upper and lower switch tubes S a and S a of the inverter A-phase, T PWM is a modulation wave period, then T PWM =T d +T s1 +T s2 .

[0061] According to the area equivalent principle, the average error voltage Δu can be obtained by averaging the error voltage in the cycle time period:

[0062]

[0063] In the formula, u dc is the DC voltage, T PWM is a modulation wave period, T d is the dead time, v t is the on voltage drop of the switch tube, and v d is the on voltage drop of the freewheeling diode.

[0064] After averaging the error voltage in the cycle time period, a square wave signal can be obtained, and Fourier decomposition of the square wave signal can obtain the mathematical expression of the average error voltage of the inverter:

[0065]

[0066] Since the permanent magnet synchronous motor is mostly connected in three-phase star, the third and integer multiple of three harmonic components will be cancelled, so only the 5th, 7th, 11th, etc. harmonic components are left in the above formula. According to the harmonic amplitude judgment, this method takes the 5th and 7th as the main research objects.

[0067] Therefore, the motor stator voltage equation in the three-phase stationary coordinate system is:

[0068]

[0069] In the formula: u1, u5 and u7 are the amplitude of the fundamental, 5th and 7th harmonic voltages respectively; θ1, θ5 and θ7 are the initial phase angles of the fundamental, 5th and 7th harmonic voltages respectively.

[0070] S3: Establish a CNN-LSTM-based harmonic compensation voltage prediction model to optimize the DQ-axis stator voltage, reduce current harmonics, and suppress vibration and torque fluctuations.

[0071] CNN, or convolutional neural network, is a kind of feedforward neural network, which is specially used to process data with grid structure. It is mainly composed of input layer, convolutional layer, pooling layer, fully connected layer and output layer. Because CNN uses convolution operation in calculation, its operation speed is greatly improved compared with general matrix operation; the alternating use of convolutional layer and pooling layer of CNN can effectively extract local features and reduce the dimension of local features.

[0072] LSTM, or long short-term memory neural network, is a kind of time recurrent neural network, which is an improvement of recurrent neural network (RNN). It effectively solves the gradient disappearance and gradient explosion of RNN by adding gate structure. The cell memory unit added in LSTM network makes LSTM network have good memory ability, and it is widely used in time series prediction.

[0073] S31: Establish a current harmonic model

[0074] The three-phase voltage u a , u b , u c of PMSM is transformed to DQ rotating coordinate system to obtain:

[0075]

[0076] In the formula, u d1 and u q1 are the d-axis and q-axis components of the fundamental voltage in the DQ rotating coordinate system, u5 and u7 are the amplitudes of the 5th and 7th harmonic voltages in the DQ rotating coordinate system, θ5 and θ7 are the initial phase angles of the 5th and 7th harmonic voltages, and ω is the rotor angular velocity.

[0077] The mathematical equation of motor current in DQ rotating coordinate system is:

[0078]

[0079] In the formula, i d1 and i q1 are the d-axis and q-axis components of the fundamental current in the DQ rotating coordinate system, i5 and i7 are the amplitudes of the 5th and 7th harmonic currents in the DQ rotating coordinate system, θ5 and θ7 are the initial phase angles of the 5th and 7th harmonic voltages, and ω is the rotor angular velocity.

[0080] The motor stator voltage equation containing harmonic components is:

[0081]

[0082] In the formula, i d1 and i q1 R represents the d-axis and q-axis components of the fundamental current in the DQ rotating coordinate system, i5 and i7 represent the amplitudes of the 5th and 7th harmonic currents in the DQ rotating coordinate system, respectively. s L is the stator resistance. d and L q θ5 and θ7 are the d-axis and q-axis inductances, respectively, θ5 and θ7 are the initial phase angles of the 5th and 7th harmonic voltages, respectively, and ω is the rotor electric angular velocity.

[0083] Since the fundamental frequency is an AC component in the DQ rotating coordinate system of the 5th and 7th harmonics, the steady-state voltage harmonic equation in the 5th DQ rotating coordinate system is:

[0084]

[0085] In the formula, i d5 and i q5 These are the DC current components along the d-axis and q-axis in a 5-order DQ rotating coordinate system, R. s L is the stator resistance. d and L q ω represents the d-axis and q-axis inductances, respectively, and ω is the rotor electrical angular velocity.

[0086] Similarly, the harmonic steady-state voltage equation in the 7th order DQ rotating coordinate system can be obtained as follows:

[0087]

[0088] In the formula, i d7 and i q7 These are the DC current components along the d-axis and q-axis in a 7th DQ rotating coordinate system, R. s L is the stator resistance. d and L q ω represents the d-axis and q-axis inductances, respectively, and ω is the rotor electrical angular velocity.

[0089] After the three-phase current of the motor passes through the 5th and 7th rotating coordinate systems of the DQ axis, the components of the 5th and 7th current harmonics on the DQ axis are obtained by low-pass filtering, such as... Figure 4 As shown.

[0090] By superimposing the harmonic voltage steady-state equation and the voltages generated in the two circuits of the PI controller, the harmonic compensation voltages of each harmonic current in the rotating coordinate system are obtained, such as... Figure 5 As shown.

[0091] Then, sum the harmonic voltages of each current to obtain the harmonic compensation voltage in the DQ coordinate system. and

[0092] That is,

[0093] S32: Collect training data

[0094] In order to ensure the accuracy of the harmonic compensation voltage prediction model, the data of the motor under each working condition is collected as much as possible, but the working conditions of the motor are countless. Therefore, in order to ensure the accuracy of the harmonic compensation voltage prediction model while improving the calculation efficiency, the initial speed of the motor is set to 50 rpm, and the torque of the motor is gradually increased from 0 Nm to the maximum torque, with an increment of 10 Nm. The data collected at each working condition is 1 second. The speed is increased to the maximum speed with an increment of 50 rpm, and the data is repeatedly recorded under each working condition. The d-axis current i d , q-axis current i q , rotor electric angle speed θ, rotor angle ω and motor reference torque Data are recorded under each working condition; then a large amount of experimental data is normalized and used to train CNN-LSTM.

[0095] S33: Establish a harmonic compensation voltage prediction model based on CNN-LSTM

[0096] The CNN (Convolutional Neural Network) is combined with the LSTM (Long Short Term Memory) to establish a CNN-LSTM network model based on Figure 6 .

[0097] The training of CNN-LSTM network is divided into two stages: the first stage, i d , i q , θ, ω and Data are input into the CNN, and the neural network is composed of four layers, namely convolution layer, Relu activation function layer, maximum pooling layer and flat layer. The second stage, the features extracted by CNN are transmitted to LSTM, and the full connection layer is used as the prediction result.

[0098] The CNN network is used to identify the initial features of the input data, and then the LSTM network is used to iteratively train the selected feature values. The final output prediction result is the harmonic compensation voltage. The mapping relationship is established with i d , i q , θ, ω and as input variables, and as output. The training process of the harmonic compensation voltage prediction model is shown in Figure 7 .

[0099] S4: Embed the harmonic compensation voltage prediction model into the motor control system to suppress the torque fluctuation of the motor.

[0100] According to the above-mentioned trained harmonic compensation voltage prediction model is embedded into the motor control system, the motor i d 、 q 、θ、ω and Data input to the prediction model of the harmonic compensation voltage, the prediction result of CNN-LSTM and Input to the control system to suppress the torque fluctuation of the motor, as Figure 8 Indicated.

[0101] The above-mentioned ideal embodiment according to the present application is inspired, through the above-mentioned description, the relevant staff can make various changes and modifications without deviating from the scope of the technical idea of the present application. The technical scope of the present application is not limited to the contents of the specification, and the technical scope must be determined according to the scope of claims.

Claims

1. A method for suppressing motor torque ripple based on CNN-LSTM, characterized in that: Includes the following steps: S1: Establish the mathematical model of the permanent magnet synchronous motor; S2: Establish an inverter model that considers the inverter dead zone and tube voltage drop; S3: Establish a harmonic compensation voltage prediction model based on CNN-LSTM to optimize the DQ axis stator voltage, reduce current harmonics, and suppress motor vibration and torque fluctuations; specifically including the following steps: S31: Establish the current harmonic model; The current harmonic model includes the steady-state voltage equations of the 5th and 7th harmonics in the DQ rotating coordinate system. The basic value of the DQ axis harmonic compensation voltage is obtained by superimposing the harmonic steady-state voltage equations with the output voltage of the PI controller. S32: Collect training data; set the initial motor speed to 50 rpm, and gradually increase the motor torque from 0 Nm to the maximum torque in increments of 10 Nm; collect data for 1 second at each operating condition; increase the speed to the maximum speed in increments of 50 rpm, and repeat the data recording under each operating condition; record the d-axis current under each operating condition. q-axis current Rotor angle Rotor electric angular velocity and motor reference torque The data was then normalized and used to train the CNN-LSTM. S33: Establish a harmonic compensation voltage prediction model based on CNN-LSTM; the specific method is as follows: By combining convolutional neural networks with long short-term neural networks, a CNN-LSTM-based network model is established; training the CNN-LSTM network involves two stages: The first phase will , , , and Data is input into the convolutional neural network; In the second stage, the features extracted by the convolutional neural network are passed to the long short-term neural network, and a fully connected layer is used as the prediction result. A convolutional neural network is used to perform initial feature recognition on the input data. Then, a long short-term neural network is used to iteratively train the selected feature values. The final output prediction result is the harmonic compensation voltage. A system is established based on... , , , and The input variable is the harmonic compensation voltage. and This represents the mapping relationship for the output; S4: Embed the CNN-LSTM-based harmonic compensation voltage prediction model from step S3 into the motor control system to suppress motor torque fluctuations.

2. The method for suppressing motor torque ripple based on CNN-LSTM according to claim 1, characterized in that: The specific method for establishing the current harmonic model in step S31 is as follows: The three-phase voltage of the permanent magnet synchronous motor , , Transforming to the DQ rotating coordinate system yields: ； In the formula, and These represent the d-axis and q-axis components of the fundamental voltage in the DQ rotating coordinate system, respectively. and The values ​​represent the amplitudes of the 5th and 7th harmonic voltages in the DQ rotating coordinate system, respectively. and The initial phase angles of the 5th and 7th harmonic voltages are respectively. The rotor's electric angular velocity; The mathematical equation for the motor current in the DQ rotating coordinate system is: ； In the formula, and These represent the d-axis and q-axis components of the fundamental current in the DQ rotating coordinate system, respectively. and The values ​​represent the amplitudes of the 5th and 7th harmonic currents in the DQ rotating coordinate system, respectively. and The initial phase angles of the 5th and 7th harmonic voltages are respectively. It is the rotor's electrical angular velocity; The stator voltage equation of the motor containing harmonic components is: ； In the formula, and These represent the d-axis and q-axis components of the fundamental current in the DQ rotating coordinate system, respectively. and The values ​​represent the amplitudes of the 5th and 7th harmonic currents in the DQ rotating coordinate system, respectively. For stator resistance, and These are the d-axis and q-axis inductances, respectively. and The initial phase angles of the 5th and 7th harmonic voltages are respectively. The rotor's electric angular velocity; Since the fundamental frequency is an AC component in the DQ rotating coordinate system for the 5th and 7th harmonics, the harmonic steady-state voltage equation in the 5th DQ rotating coordinate system is: ； In the formula, and These are the DC current components along the d-axis and q-axis under the fifth DQ rotating coordinate system. For stator resistance, and These are the d-axis and q-axis inductances, respectively. The rotor's electric angular velocity; Similarly, the harmonic steady-state voltage equation in the 7th order DQ rotating coordinate system can be obtained as follows: ； In the formula, and These represent the DC current components along the d-axis and q-axis under the 7th DQ rotating coordinate system. For stator resistance, and These are the d-axis and q-axis inductances, respectively. The rotor's electric angular velocity; After the three-phase current of the motor passes through the 5th and 7th rotating coordinate systems of the DQ axis, the components of the 5th and 7th current harmonics on the DQ axis are obtained by low-pass filtering. By superimposing the harmonic steady-state voltage equation and the voltages generated in the two circuits of the PI controller, the harmonic compensation voltages of each harmonic current in the rotating coordinate system are obtained. Then, sum the harmonic voltages of each current to obtain the harmonic compensation voltage in the DQ coordinate system. and ,Right now: .

Images