Motor vector control method based on high-frequency signal injection and related equipment

Abstract

The invention provides a motor vector control method based on high-frequency signal injection and related equipment. The method comprises the following steps: adjusting an injection parameter of a high-frequency signal according to a motor operation state; determining a high-frequency current response signal of the motor according to the injection parameter; extracting a rotor position estimation value corresponding to the high-frequency current response signal through a deep learning model; inputting the rotor position estimation value into a motor dynamic model to generate a joint estimation value of the rotor position and the speed; and feeding back the joint estimation value to a motor control loop to carry out motor vector control. Through the implementation of the scheme, the rotor position estimation value corresponding to the high-frequency current response signal is extracted through the deep learning model, the accuracy and robustness of rotor position estimation can be improved, and estimation errors caused by motor parameter changes or external disturbance are reduced.

Description

Technical Field

[0001] The present application relates to the field of motor control, and in particular to a motor vector control method based on high-frequency signal injection and related equipment. Background Art

[0002] Performance Motor control has a wide range of applications in industrial automation, new energy vehicles, aerospace, and intelligent manufacturing, especially sensorless control technology, which can reduce system costs, improve reliability, and reduce maintenance requirements. The high-frequency signal injection (HFI) method is a common sensorless control strategy that is suitable for scenarios where traditional back-EMF methods fail during low-speed or zero-speed operation. By injecting high-frequency signals into the stator windings, the system can use the changes in the internal magnetic resistance characteristics of the motor to estimate the rotor position and speed. However, traditional HFI methods rely on signal processing and observers based on mathematical models (such as phase-locked loops, extended Kalman filters, sliding mode observers, etc.), which have certain limitations in high-frequency signal demodulation, noise suppression, and nonlinear modeling. Especially under complex operating conditions, such as load changes, motor parameter drift, and enhanced electromagnetic interference, the robustness of traditional methods decreases, resulting in reduced accuracy in rotor position and speed estimation. Summary of the invention

[0003] The present application provides a motor vector control method based on high-frequency signal injection and related equipment, which are used to solve the problem of low accuracy of rotor position and speed estimation under complex operating conditions in the related art.

[0004] A first aspect of the present application provides a motor vector control method based on high-frequency signal injection, the motor vector control method based on high-frequency signal injection comprising: Adjust the injection parameters of the high-frequency signal according to the motor operating status; Determining a high-frequency current response signal of the motor according to the injection parameter; Extracting a rotor position estimation value corresponding to the high-frequency current response signal through a deep learning model; Inputting the rotor position estimate into a motor dynamics model to generate a joint estimate of the rotor position and speed; The combined estimated value is fed back to the motor control loop to perform motor vector control.

[0005] Optionally, in a first implementation of the first aspect of the present application, the injection parameter includes an amplitude and a frequency of the high-frequency signal, and the step of adjusting the injection parameter of the high-frequency signal according to the running state of the motor includes: Determine the initial amplitude and initial frequency of the high-frequency signal according to the motor operation state; Monitor the high-frequency current response signal of the motor in real time to determine the injection effect of the high-frequency signal; The amplitude and frequency of the high-frequency signal are dynamically adjusted according to the judgment result of the high-frequency current response signal.

[0006] Optionally, in a second implementation of the first aspect of the present application, the step of extracting the rotor position estimation value corresponding to the high-frequency current response signal through a deep learning model includes: Preprocessing the high-frequency current response signal; Segmenting the preprocessed high-frequency current response signal according to predefined time windows; Inputting the segmented signal into the deep learning model to extract rotor position features; A rotor position estimation value corresponding to the high-frequency current response signal is determined according to the rotor position characteristic.

[0007] Optionally, in a third implementation of the first aspect of the present application, the step of inputting the estimated value of the rotor position into a motor dynamics model to generate a joint estimated value of the rotor position and speed includes: Determine the first rotor position and velocity and covariance matrix at an initial time step; Predicting a second rotor position and speed in a next time step of the initial time step according to a motor dynamics model; Updating the covariance matrix according to the Jacobian matrix of the motor dynamics model; Calculate the Kalman gain based on the updated covariance matrix; A joint estimate of rotor position and speed is determined based on the rotor position estimate and the Kalman gain.

[0008] Optionally, in a fourth implementation of the first aspect of the present application, after the step of determining the high-frequency current response signal of the motor according to the injection parameter, the step further includes: Predicting a current variation trend according to the high-frequency current response signal; When it is detected that the current reaches a current threshold, adjusting the injection parameters of the high-frequency signal; And / or, dynamically adjusting the distribution ratio of the direct-axis current and the quadrature-axis current according to the motor operating state and control target.

[0009] Optionally, in a fifth implementation of the first aspect of the present application, the method further includes: Allocating different high-frequency signal frequencies or high-frequency signal injection time periods to N motors; wherein N is an integer greater than or equal to 1; Dynamically adjust the frequency of the high-frequency signal or the allocation strategy of the high-frequency signal injection period according to the operating state of the motor; Injecting the high frequency signal into the N motors according to the allocation strategy; Collecting mixed signals of the N motors through current sensors; The high-frequency current response signal of each motor is separated from the mixed signal.

[0010] Optionally, in a sixth implementation of the first aspect of the present application, the method further includes: Identify the motor fault type by monitoring the abnormality of the high-frequency current response signal; generating a fault-tolerant control signal according to the motor fault type; The injection parameter of the high frequency signal is adjusted according to the fault-tolerant control signal.

[0011] A second aspect of the present application provides a motor vector control device based on high-frequency signal injection, the motor vector control device based on high-frequency signal injection comprising: An adjustment module, used for adjusting injection parameters of the high-frequency signal according to the running state of the motor; A determination module, used to determine a high-frequency current response signal of the motor according to the injection parameter; An extraction module, used to extract a rotor position estimation value corresponding to the high-frequency current response signal through a deep learning model; A generating module, used for inputting the estimated value of the rotor position into a motor dynamics model to generate a joint estimated value of the rotor position and speed; A control module is used to feed back the joint estimation value to a motor control loop to perform motor vector control.

[0012] A third aspect of an embodiment of the present application provides an electronic device, comprising a memory and a processor, wherein the processor is used to execute a computer program stored in the memory, and when the processor executes the computer program, it implements each step of the motor vector control method based on high-frequency signal injection provided in the first aspect of the embodiment of the present application.

[0013] The fourth aspect of the embodiments of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, each step of the motor vector control method based on high-frequency signal injection provided in the first aspect of the embodiments of the present application is implemented.

[0014] In summary, according to the motor vector control method and related equipment based on high-frequency signal injection provided by the present application, the injection parameters of the high-frequency signal are adjusted according to the motor operating state; the high-frequency current response signal of the motor is determined according to the injection parameters; the rotor position estimate corresponding to the high-frequency current response signal is extracted through a deep learning model; the rotor position estimate is input into the motor dynamics model to generate a joint estimate of the rotor position and speed; the joint estimate is fed back to the motor control loop for motor vector control. Through the implementation of the present application, the rotor position estimate corresponding to the high-frequency current response signal is extracted through a deep learning model, which can improve the accuracy and robustness of the rotor position estimation and reduce the estimation error caused by motor parameter changes or external disturbances. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic flow chart of a motor vector control method based on high-frequency signal injection provided in an embodiment of the present application; Figure 2 A schematic diagram of a program module of a motor vector control device based on high-frequency signal injection provided in an embodiment of the present application; Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] In order to make the purpose, features, and advantages of the invention of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0017] In order to solve the problem of low accuracy of rotor position and speed estimation under complex operating conditions in the related art, the embodiment of the present application provides a motor vector control method based on high-frequency signal injection, such as Figure 1 The flowchart of the motor vector control method based on high-frequency signal injection provided in this embodiment is as follows. The motor vector control method based on high-frequency signal injection includes the following steps: Step 110: Adjust the injection parameters of the high-frequency signal according to the operating state of the motor.

[0018] Specifically, the motor operating state includes parameters such as speed, load, and temperature. These parameters directly affect the dynamic response of the motor and the injection effect of the high-frequency signal. Real-time monitoring of these parameters by sensors can provide a basis for the adjustment of the high-frequency signal. The injection parameters of the high-frequency signal include the amplitude and frequency of the high-frequency signal. The amplitude and frequency of the high-frequency signal need to be dynamically adjusted according to the operating state of the motor. For example, when running at a low speed, a higher signal amplitude may be required to enhance the detectability of the signal. In this embodiment, a speed sensor, a temperature sensor, and a current sensor are used to collect the operating state data of the motor in real time, and the collected data is transmitted to the control system for processing. According to the collected data, the preset parameter adjustment rules or models are queried to determine the optimal high-frequency signal amplitude and frequency in the current state, and the adjusted high-frequency signal is injected into the stator winding of the motor.

[0019] In an optional implementation of the present embodiment, the step of adjusting the injection parameters of the high-frequency signal according to the operating state of the motor includes: determining the initial amplitude and initial frequency of the high-frequency signal according to the operating state of the motor; monitoring the high-frequency current response signal of the motor in real time to judge the injection effect of the high-frequency signal; and dynamically adjusting the amplitude and frequency of the high-frequency signal according to the judgment result of the high-frequency current response signal.

[0020] Specifically, in this embodiment, in the motor vector control, the injection parameters of the high-frequency signal need to be dynamically adjusted according to the real-time operating state of the motor to ensure the effectiveness of the signal and the stability of the system. First, the initial amplitude and frequency of the high-frequency signal need to be determined by comprehensively considering the key parameters of the motor such as the speed, load, and temperature. For example, when the motor is running at a low speed, the rotor position information is easily interfered by noise. At this time, it is necessary to appropriately increase the amplitude of the high-frequency signal to enhance the detectability of the signal; while in a high temperature environment, the impedance characteristics of the motor winding may change, and the signal frequency needs to be reduced to avoid signal distortion caused by thermal effects. After the high-frequency signal is injected, by installing a current sensor and a data acquisition system, the control system can obtain the current changes in real time, especially the current response caused by the high-frequency signal. These signals are usually manifested as high-frequency components superimposed on the original current signal. The amplitude, phase, and frequency characteristics of the current response can reflect the operating state of the motor and the injection effect of the signal. The control system will analyze these response signals in real time to identify whether the high-frequency signal has achieved the expected effect. If the amplitude change and frequency characteristics of the response signal meet the preset requirements, the system can consider that the signal injection effect is good and no adjustment is required; if the change of the response signal is small or deviates from the expected range, the system needs to adjust the signal injection parameters. For example, when the motor load is large, the amplitude of the current response may be suppressed. At this time, if the amplitude of the injected signal is too low, it may not be able to effectively stimulate the high-frequency response of the motor. To this end, the system will increase the amplitude of the signal through a feedback mechanism to increase the influence of the signal. If the frequency of the signal is too low, the high-frequency response of the motor may lag when running at high speed, thereby affecting the accuracy of the rotor position estimation. At this time, the system will appropriately increase the frequency of the signal to ensure that the dynamic characteristics of the motor at high-speed rotation can be captured.

[0021] Step 120: Determine the high-frequency current response signal of the motor according to the injection parameters.

[0022] Specifically, in this embodiment, after the high-frequency signal is injected into the motor, the motor winding will generate a corresponding high-frequency current response signal, which contains the rotor position information. However, due to noise and nonlinear factors, further processing is required. The collected signal is transmitted to the signal processing unit, and the signal is pre-processed, including but not limited to low-pass filtering the signal, removing high-frequency noise, normalizing the signal, and eliminating amplitude differences to improve signal quality. The features of the high-frequency current response signal are extracted through signal processing technology (such as Fourier transform and wavelet transform), and these features will be used for subsequent rotor position estimation.

[0023] In an optional implementation of the present embodiment, the step of extracting the rotor position estimate corresponding to the high-frequency current response signal through a deep learning model includes: preprocessing the high-frequency current response signal; segmenting the preprocessed high-frequency current response signal according to a predefined time window; inputting the segmented signal into the deep learning model to extract the rotor position features; and determining the rotor position estimate corresponding to the high-frequency current response signal based on the rotor position features.

[0024] Specifically, in this embodiment, the collected high-frequency current response signal needs to be preprocessed first to remove noise and interference in the signal and improve the quality and reliability of the signal. The preprocessing method includes but is not limited to filtering, denoising and normalization. After the preprocessing is completed, the preprocessed high-frequency current response signal needs to be segmented according to a predefined time window, and the continuous signal is divided into multiple fixed-length segments so that each segment contains enough information to reflect the operating state of the motor. The length of the time window should be selected according to the operating characteristics of the motor and the frequency characteristics of the signal, and the time resolution and frequency resolution of the signal need to be balanced. The segmented signal segments are obtained by sliding windows to ensure that there is a certain overlap between the signal segments, thereby avoiding information loss. The segmented signal segments are input into the trained deep learning model, and the rotor position features are extracted by forward calculation. The output of the model is a feature vector, which represents the various feature information contained in the signal segment, including the amplitude, frequency, phase, etc. of the signal, reflecting the dynamic characteristics of the signal segment and the change of the rotor position. According to these extracted feature information, the feature vector is mapped to the rotor position estimate through the regression model.

[0025] It should be noted that the goal of the convolutional neural network is to extract rotor position features from the high-frequency current response signal. This feature extraction process is completed through multiple convolutional layers, pooling layers, and fully connected layers. The purpose of CNN training is to make the mapping from input signals to rotor position features more accurate by optimizing the loss function. The training process achieves this goal by adjusting the weights and biases in the network. Let the high-frequency current response signal be X, and its shape is [T, N], where T is the time length of the signal and N is the number of sampling points per time step. In the convolution layer, the signal is convolved through a filter (convolution kernel) to output a feature map. for: , in, is the convolution kernel, The convolution operation extracts features from the signal by sliding the convolution kernel and calculating the dot product. The activation function is used to introduce nonlinearity so that the neural network can learn more complex patterns. Activation function: , The pooling layer is used to reduce the size of the feature map and highlight the most significant features. ) operation is used to extract the strongest local features in the feature map: , The pooled feature map is passed to the fully connected layer for further feature extraction. The output of the fully connected layer is: , in, is the weight matrix, Also the bias, the loss function used in the training process is: , in, is the actual rotor position label, To obtain the rotor position features predicted by CNN, in order to optimize the network parameters by minimizing the loss function in each iteration, the gradient descent method (or other optimization algorithms) is used to update the weights in the network: , , in, is the learning rate, , The loss function is the weight as well as The above formula can be understood as the weight of the nth iteration is the weight obtained after the previous iteration learning. Through the back propagation algorithm, the weight of the network is continuously adjusted so that the network can learn more accurate rotor position characteristics from the input signal.

[0026] After extracting the rotor position features, the next task is to convert these features into specific rotor position estimates through a regression model. The purpose of the regression model is to output an accurate rotor position estimate by learning the mapping relationship between the feature vector and the rotor position. Use the regression model to map the feature vector to the rotor position estimate superior: , in, is the weight of the regression model, is the bias term. Similarly, the regression model is iterated using the loss function and weight update method to finally generate an accurate rotor position estimate.

[0027] Step 130: extracting a rotor position estimation value corresponding to the high-frequency current response signal through a deep learning model.

[0028] Specifically, a deep learning model suitable for signal processing is selected, such as a convolutional neural network (CNN) or a long short-term memory network (LSTM), a data set containing a high-frequency current response signal and a rotor position label is constructed, the data set is used to train the deep learning model, the model parameters are optimized, the high-frequency current response signal collected in real time is input into the trained deep learning model, and the model outputs an estimated value of the rotor position. In this embodiment, high-frequency current response signals are collected under different working conditions, and the corresponding rotor position information is recorded. 80% of the data set is used to train the deep learning model, and 20% of the data set is used for testing. The high-frequency current response signal collected in real time is input into the trained deep learning model, and the model outputs an estimated value of the rotor position.

[0029] In an optional implementation of the present embodiment, the step of inputting the rotor position estimate into the motor dynamics model to generate a joint estimate of the rotor position and speed includes: determining the first rotor position and speed and the covariance matrix of an initial time step; predicting the second rotor position and speed of the next time step of the initial time step according to the motor dynamics model; updating the covariance matrix according to the Jacobian matrix of the motor dynamics model; calculating the Kalman gain according to the updated covariance matrix; and determining the joint estimate of the rotor position and speed according to the rotor position estimate and the Kalman gain.

[0030] Specifically, in this embodiment, in the process of realizing the joint estimation of the motor rotor position and speed, it is first necessary to determine the first rotor position and speed and the covariance matrix at the initial time step. and speed It can be obtained through sensor measurement or system initial setting. Initial covariance matrix Used to represent the uncertainty of the initial state estimate. It is a symmetric positive definite matrix, usually set empirically or obtained through prior knowledge. The diagonal elements of the initial covariance matrix represent the variance of the state variables, and the off-diagonal elements represent the covariance between the state variables. The motor dynamics model describes the changing relationship between the rotor position and speed, and is usually composed of the rotation equation and the motion equation. For example, the rotation equation can be expressed as: The equation of motion is expressed as: , in, is the moment of inertia, is the rotor speed, is the electromagnetic torque, is the load torque, is the rotor position. The continuous equation is discretized by numerical integration methods (such as Euler's method) to predict the rotor position and speed at the next time step. For example, the discretized motion equation is: , The discretized rotation equation is: , in, is the time step. After predicting the rotor position and speed of the next time step, the covariance matrix needs to be updated. The update formula of the covariance matrix is: , in, is the prediction covariance matrix, is the Jacobian matrix of the state transfer equation, is the current state covariance matrix, is the process noise covariance matrix, which represents the noise introduced by the inherent uncertainty of the system. After updating the covariance matrix, the Kalman gain is calculated according to the updated covariance matrix. The Kalman gain is a key parameter for balancing the weight between the predicted value and the measured value. The Kalman gain is calculated by combining the predicted covariance matrix and the measured noise covariance matrix. The formula is: , in, is the Jacobian matrix of the observation matrix, describing the relationship between the measured variables and the system state variables, is the measurement noise covariance matrix. Finally, the joint estimate of the rotor position and speed is determined based on the rotor position estimate and the Kalman gain. Through the update step of the Kalman filter, the predicted value and the measured value are combined to obtain the optimal state estimate. The update formula is: , in, is the updated state estimate, including rotor position and speed, is the estimated value of the predicted state, is the measured value, For the predicted measured values, the updated covariance matrix is: , in, is the identity matrix, is the updated covariance matrix. The size of the covariance matrix reflects the credibility of the state estimation. If the value in the covariance matrix is ​​small, it means that the credibility of the estimated value is high and the error is small. Conversely, if the value in the covariance matrix is ​​large, it means that the credibility of the estimated value is low and the error is large. Therefore, by monitoring the changes in the covariance matrix, the credibility of the state estimation can be dynamically evaluated. The updating and monitoring of the covariance matrix can improve the robustness of the state estimation. In practical applications, the system may be affected by various noises and uncertainties. By dynamically updating the covariance matrix, the weight of the state estimation can be adjusted in time, and the system's adaptability to noise and uncertainties can be enhanced, thereby improving the accuracy and stability of the state estimation.

[0031] Step 140: Input the estimated value of the rotor position into the motor dynamics model to generate a joint estimated value of the rotor position and speed.

[0032] Specifically, the motor dynamics model describes the motion characteristics of the motor, including the position and speed of the rotor, and the rotor position estimate is input into the motor dynamics model, combined with historical data, to generate a joint estimate of the rotor position and speed. In this embodiment, the rotor position and speed of the initial time step and the covariance matrix are determined, the rotor position and speed of the next time step are predicted according to the motor dynamics model, and the joint estimate of the rotor position and speed is determined according to the rotor position estimate and the Kalman gain.

[0033] Step 150 : Feedback the combined estimated value to the motor control loop to perform motor vector control.

[0034] Specifically, the motor control loop includes a current loop, a speed loop, and a position loop. The joint estimate of the rotor position and speed is fed back to the control loop, the motor control signal is adjusted, and the direct axis (direct axis) and quadrature axis (quadrature axis) components of the current are adjusted according to the joint estimate to achieve precise magnetic field oriented control. Among them, the direct axis current controls the magnetic flux (magnetic flux), affecting the magnetization characteristics of the motor, and the quadrature axis current controls the electromagnetic torque, which directly determines the output torque of the motor. When running at low speed, the quadrature axis current may need to be increased to provide sufficient torque, and the direct axis current can be adjusted appropriately as needed; when running at high speed, in order to maintain efficient operation, the system will reduce the direct axis current and adjust the quadrature axis current according to the speed to ensure that the motor can maintain a stable torque output.

[0035] In an optional implementation of the present embodiment, after the step of determining the high-frequency current response signal of the motor according to the injection parameters, it also includes: predicting the current change trend according to the high-frequency current response signal; when it is detected that the current reaches the current threshold, adjusting the injection parameters of the high-frequency signal; and / or, dynamically adjusting the distribution ratio of the direct-axis and quadrature-axis currents according to the motor operating state and control objectives.

[0036] Specifically, in this embodiment, in order to predict the current change trend, it is necessary to process and analyze the collected current signal, extract the features in the current signal to model and predict the current change trend. In the process of predicting the current change trend, it is first necessary to preprocess the input signal, perform discrete Fourier transform (DFT) on the signal, and convert it into the frequency domain: , in, is the current signal, t represents the time, is the frequency domain signal, Indicates frequency, Denotes Fourier transform. Afterwards, the main frequency components and amplitudes are extracted from the frequency domain signal: , in, represents the kth main frequency component and its corresponding amplitude, M is the number of main frequency components, and the amplitude of each main frequency component is modeled and predicted using the autoregressive moving average (ARMA) model: , in, is the autoregression coefficient, is the coefficient of the sliding average part, The noise terms p and q are the orders of the autoregressive and sliding average parts, respectively. The amplitude of each main frequency component is predicted to obtain the predicted amplitude at the future moment: , Convert the predicted main frequency components and amplitudes back to the time domain to obtain the predicted current signal: , in, is the phase of the kth frequency component.

[0037] It should be noted that in the above formula, the current signal The frequency domain signal is converted to the frequency domain through discrete Fourier transform (DFT) Fourier transform is used to analyze the frequency components of a signal and identify the main frequency components and their corresponding amplitudes. Extracting the main frequency components from a frequency domain signal , these frequency components reflect the main change trend of the current signal. The extracted frequency components and amplitudes are used for subsequent time series modeling and prediction. Perform time series modeling and use the autoregressive moving average (ARMA) model for forecasting. The ARMA model combines the autoregressive (AR) and moving average (MA) parts to effectively capture trends and periodicity in time series. The amplitude of the main frequency components is predicted through the ARMA model to obtain the predicted amplitude at future moments. . Convert the predicted frequency components and amplitudes back to the time domain and reconstruct the predicted current signal The reconstruction formula combines the amplitude, frequency and phase of the main frequency components to generate the predicted value of the current signal at a future time.

[0038] When the current reaches the preset current threshold, the system needs to adjust the injection parameters of the high-frequency signal in time to avoid current overload or other abnormal conditions. The setting of the current threshold is usually determined according to the rated current and safe operating range of the motor. When it is detected that the current reaches the threshold, the impact on the motor current can be reduced by reducing the amplitude or frequency of the high-frequency signal. For example, assuming that the rated current of the motor is 10A and the current threshold is set to 9A, when the current is close to or exceeds 9A in real time, the system can reduce the amplitude of the high-frequency signal from 5V to 3V, or reduce the injection frequency from 10kHz to 5kHz, thereby reducing the current fluctuation and ensuring that the motor operates within a safe range. In addition to adjusting the injection parameters of the high-frequency signal, dynamically adjusting the distribution ratio of the direct axis (d-axis) and quadrature axis (q-axis) current according to the motor operating state and control objectives is also an important means to achieve precise control. In motor vector control, the d-axis current is mainly used to control the magnetic flux, and the q-axis current is mainly used to control the torque. According to different operating states and control objectives, the reasonable allocation of d-axis and q-axis currents can optimize the performance of the motor.

[0039] In an optional implementation of the present embodiment, different high-frequency signal frequencies or high-frequency signal injection time periods are allocated to N motors; wherein N is an integer greater than or equal to 1; the allocation strategy of the high-frequency signal frequency or the high-frequency signal injection time period is dynamically adjusted according to the operating status of the motor; high-frequency signals are injected into the N motors according to the allocation strategy; mixed signals of the N motors are collected by current sensors; and the high-frequency current response signals of each motor are separated from the mixed signals.

[0040] Specifically, in this embodiment, for each motor, the system can pre-set a set of available high-frequency signal frequencies or injection periods, and for the allocation of injection periods, the entire time period can be divided into multiple non-overlapping time windows, such as each motor injecting a high-frequency signal within a specific time window. In the initial configuration stage, the system needs to reasonably allocate these frequencies or injection periods according to the operating state, load conditions and control objectives of the motor. In this way, each motor injects a high-frequency signal within a specific frequency or time window to ensure the independence between signals. Afterwards, the allocation strategy of the high-frequency signal frequency or high-frequency signal injection period is dynamically adjusted according to the operating state of the motor. For example, when the load of a motor increases and a higher frequency signal is required to ensure accurate rotor position estimation, the system can increase the high-frequency signal frequency of the motor and adjust the frequencies of other motors at the same time to avoid signal conflicts. Similarly, when the operating state of a motor is relatively stable, its high-frequency signal injection period can be reduced, and more time windows can be allocated to other motors to optimize signal utilization efficiency. After completing the allocation and adjustment of the high-frequency signal, the system injects high-frequency signals into N motors according to the current allocation strategy. Each motor receives the high-frequency signal within its specific frequency or time window. The mixed signal of the N motors is collected by the current sensor. The system needs to install a sensor in the current loop for real-time monitoring and collection of the current response signal of the motor. Since the high-frequency signals of multiple motors act in the current loop at the same time, the signal collected by the current sensor is the superposition of the high-frequency current response signals of multiple motors. In the signal separation process, the high-frequency signal frequency or injection period previously allocated to each motor can be used as the basis for signal separation. For example, for frequency allocation, the components of different frequencies can be separated by a bandpass filter, and each bandpass filter corresponds to the high-frequency signal frequency of a motor, thereby extracting the high-frequency current response signal of the motor. For injection period allocation, the mixed signal can be segmented according to a pre-set time window through a time domain window function, and each time window corresponds to the high-frequency signal injection period of a motor, thereby separating the high-frequency current response signal of the motor.

[0041] In an optional implementation of the present embodiment, the motor fault type is identified by monitoring the abnormality of the high-frequency current response signal; a fault-tolerant control signal is generated according to the motor fault type; and an injection parameter of the high-frequency signal is adjusted according to the fault-tolerant control signal.

[0042] Specifically, in this embodiment, the high-frequency current response signal of the motor is monitored in real time by sensors and data acquisition systems. The high-frequency current response signal contains important information about the motor's operating status. When the motor fails, the characteristics of these signals will change. For example, when the motor has a winding short circuit, the amplitude and frequency components of the high-frequency current response signal may change significantly; when the motor has a rotor imbalance, certain frequency components in the signal will increase or decrease. By continuously collecting these high-frequency signals, the system can obtain the current response data at each moment during the operation of the motor. After the high-frequency current response signal is collected, the signal features are extracted using signal processing and analysis techniques. When the system detects abnormal features in the high-frequency current response signal and determines the fault type of the motor through a fault identification model, the system needs to generate a corresponding fault-tolerant control signal. The fault-tolerant control signal is used to adjust the operating parameters of the motor to ensure that the motor can still operate stably when a fault occurs, avoiding serious damage or shutdown. For example, when a winding short-circuit fault is detected, the system can generate a fault-tolerant control signal that reduces the current amplitude to reduce the risk of winding overheating and damage; when a rotor imbalance fault is detected, the system can generate a fault-tolerant control signal that adjusts the torque output to balance the rotor and reduce vibration. Through the generated fault-tolerant control signal, the system can dynamically adjust the injection parameters of the high-frequency signal to adapt to the fault state of the motor. For example, when a winding short-circuit fault is detected, in order to avoid current overload, the system can reduce the amplitude of the high-frequency signal and reduce the impact on the motor current; when a rotor imbalance fault is detected, in order to balance the rotor and reduce vibration, the system can adjust the frequency of the high-frequency signal to avoid the resonant frequency.

[0043] According to a motor vector control method based on high-frequency signal injection provided by the present application, the injection parameters of the high-frequency signal are adjusted according to the motor operating state; the high-frequency current response signal of the motor is determined according to the injection parameters; the rotor position estimate corresponding to the high-frequency current response signal is extracted through a deep learning model; the rotor position estimate is input into the motor dynamics model to generate a joint estimate of the rotor position and speed; the joint estimate is fed back to the motor control loop for motor vector control. Through the implementation of the present application, the rotor position estimate corresponding to the high-frequency current response signal is extracted through a deep learning model, which can improve the accuracy and robustness of the rotor position estimation and reduce the estimation error caused by motor parameter changes or external disturbances.

[0044] Figure 2 A motor vector control device based on high-frequency signal injection is provided in an embodiment of the present application. The motor vector control device based on high-frequency signal injection can be used to implement the motor vector control method based on high-frequency signal injection in the above-mentioned embodiment. Figure 2 As shown, the motor vector control device based on high-frequency signal injection mainly includes: An adjustment module 10, used to adjust the injection parameters of the high-frequency signal according to the running state of the motor; A determination module 20, configured to determine a high-frequency current response signal of the motor according to the injection parameters; An extraction module 30, used to extract a rotor position estimation value corresponding to the high-frequency current response signal through a deep learning model; A generating module 40, for inputting the estimated value of the rotor position into the motor dynamics model to generate a joint estimated value of the rotor position and speed; The control module 50 is used to feed back the joint estimation value to the motor control loop to perform motor vector control.

[0045] In an optional implementation of the present embodiment, the adjustment module is specifically used to: determine the initial amplitude and initial frequency of the high-frequency signal according to the operating state of the motor; monitor the high-frequency current response signal of the motor in real time to judge the injection effect of the high-frequency signal; and dynamically adjust the amplitude and frequency of the high-frequency signal according to the judgment result of the high-frequency current response signal.

[0046] In an optional implementation of the present embodiment, the extraction module is specifically used to: preprocess the high-frequency current response signal; segment the preprocessed high-frequency current response signal according to a predefined time window; input the segmented signal into a deep learning model to extract rotor position features; and determine the rotor position estimate corresponding to the high-frequency current response signal based on the rotor position features.

[0047] In an optional implementation of the present embodiment, the generation module is specifically used to: determine the first rotor position and speed and the covariance matrix of an initial time step; predict the second rotor position and speed of the next time step of the initial time step according to the motor dynamics model; update the covariance matrix according to the Jacobian matrix of the motor dynamics model; calculate the Kalman gain according to the updated covariance matrix; and determine a joint estimate of the rotor position and speed according to the rotor position estimate and the Kalman gain.

[0048] In an optional implementation of this embodiment, the motor vector control device further includes: a prediction module. The prediction module is used to: predict the current change trend according to the high-frequency current response signal. The adjustment module is also used to: when it is detected that the current reaches the current threshold, adjust the injection parameter of the high-frequency signal; and / or, according to the motor operating state and control target, dynamically adjust the distribution ratio of the direct-axis and quadrature-axis currents.

[0049] In an optional implementation of the present embodiment, the determination module is also used to: assign different high-frequency signal frequencies or high-frequency signal injection time periods to N motors; wherein N is an integer greater than or equal to 1; dynamically adjust the allocation strategy of the high-frequency signal frequency or the high-frequency signal injection time period according to the operating status of the motor; inject high-frequency signals into the N motors according to the allocation strategy; collect mixed signals of the N motors through current sensors; and separate the high-frequency current response signals of each motor from the mixed signals.

[0050] In an optional implementation of the present embodiment, the adjustment module is also used to: identify the motor fault type by monitoring the abnormality of the high-frequency current response signal; generate a fault-tolerant control signal according to the motor fault type; and adjust the injection parameters of the high-frequency signal according to the fault-tolerant control signal.

[0051] According to a motor vector control device based on high-frequency signal injection provided by the present application, the injection parameters of the high-frequency signal are adjusted according to the motor operating state; the high-frequency current response signal of the motor is determined according to the injection parameters; the rotor position estimate corresponding to the high-frequency current response signal is extracted through a deep learning model; the rotor position estimate is input into the motor dynamics model to generate a joint estimate of the rotor position and speed; the joint estimate is fed back to the motor control loop to perform motor vector control. Through the implementation of the present application, the rotor position estimate corresponding to the high-frequency current response signal is extracted through a deep learning model, which can improve the accuracy and robustness of the rotor position estimation and reduce the estimation error caused by motor parameter changes or external disturbances.

[0052] According to the application plan provided Figure 3 An electronic device is provided in an embodiment of the present application. The electronic device can be used to implement the motor vector control method based on high-frequency signal injection in the above-mentioned embodiment, mainly including: The memory 301, the processor 302 and the computer program 303 stored in the memory 301 and executable on the processor 302, the memory 301 and the processor 302 are connected by communication. When the processor 302 executes the computer program 303, the motor vector control method based on high-frequency signal injection in the above-mentioned embodiment is implemented. The number of processors can be one or more.

[0053] The memory 301 may be a high-speed random access memory (RAM) memory, or a non-volatile memory (non-volatile memory), such as a disk memory. The memory 301 is used to store executable program codes, and the processor 302 is coupled to the memory 301 .

[0054] Furthermore, the present application also provides a computer-readable storage medium, which may be provided in the electronic device in the above embodiments. Figure 3 Memory in the illustrated embodiment.

[0055] The computer readable storage medium stores a computer program, and when the program is executed by the processor, the motor vector control method based on high-frequency signal injection in the aforementioned embodiment is implemented. Furthermore, the computer storable medium can also be a U disk, a mobile hard disk, a read-only memory (ROM), a RAM, a magnetic disk or an optical disk, and other media that can store program codes.

[0056] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0057] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0058] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A motor vector control method based on high-frequency signal injection, characterized in that: include: Adjust the injection parameters of the high-frequency signal according to the motor operating status; Determining a high-frequency current response signal of the motor according to the injection parameter; Extracting a rotor position estimation value corresponding to the high-frequency current response signal through a deep learning model; Inputting the rotor position estimate into a motor dynamics model to generate a joint estimate of the rotor position and speed; Feeding back the combined estimated value to a motor control loop to perform motor vector control; The step of inputting the estimated value of the rotor position into the motor dynamics model to generate a joint estimated value of the rotor position and speed comprises: Determine the first rotor position and velocity and covariance matrix at an initial time step; Predicting a second rotor position and speed in a next time step of the initial time step according to a motor dynamics model; Updating the covariance matrix according to the Jacobian matrix of the motor dynamics model; Calculate the Kalman gain based on the updated covariance matrix; A joint estimate of rotor position and speed is determined based on the rotor position estimate and the Kalman gain.

2. The motor vector control method based on high frequency signal injection according to claim 1, characterized in that: The injection parameters include the amplitude and frequency of the high-frequency signal, and the step of adjusting the injection parameters of the high-frequency signal according to the motor operation state includes: Determine the initial amplitude and initial frequency of the high-frequency signal according to the motor operation state; Monitor the high-frequency current response signal of the motor in real time to determine the injection effect of the high-frequency signal; The amplitude and frequency of the high-frequency signal are dynamically adjusted according to the judgment result of the high-frequency current response signal.

3. The motor vector control method based on high frequency signal injection according to claim 1, characterized in that: The step of extracting the rotor position estimation value corresponding to the high-frequency current response signal through the deep learning model includes: Preprocessing the high-frequency current response signal; Segmenting the preprocessed high-frequency current response signal according to predefined time windows; Inputting the segmented signal into the deep learning model to extract rotor position features; A rotor position estimation value corresponding to the high-frequency current response signal is determined according to the rotor position characteristic.

4. The motor vector control method based on high frequency signal injection according to claim 1, characterized in that: After the step of determining the high-frequency current response signal of the motor according to the injection parameter, the method further includes: predicting a current variation trend according to the high-frequency current response signal; When it is detected that the current reaches a current threshold, adjusting the injection parameters of the high-frequency signal; And / or, dynamically adjusting the distribution ratio of the direct-axis current and the quadrature-axis current according to the motor operating state and control target.

5. The motor vector control method based on high frequency signal injection according to claim 1, characterized in that: The method further comprises: Allocating different high-frequency signal frequencies or high-frequency signal injection time periods to N motors; wherein N is an integer greater than or equal to 1; Dynamically adjust the frequency of the high-frequency signal or the allocation strategy of the high-frequency signal injection period according to the operating state of the motor; Injecting the high frequency signal into the N motors according to the allocation strategy; Collecting mixed signals of the N motors through current sensors; The high-frequency current response signal of each motor is separated from the mixed signal.

6. The motor vector control method based on high frequency signal injection according to claim 1, characterized in that: The method further comprises: Identify the motor fault type by monitoring the abnormality of the high-frequency current response signal; generating a fault-tolerant control signal according to the motor fault type; The injection parameter of the high frequency signal is adjusted according to the fault-tolerant control signal.

7. A motor vector control device based on high-frequency signal injection, characterized in that: The motor vector control device based on high-frequency signal injection includes: An adjustment module, used for adjusting injection parameters of the high-frequency signal according to the running state of the motor; A determination module, used to determine a high-frequency current response signal of the motor according to the injection parameter; An extraction module, used to extract a rotor position estimation value corresponding to the high-frequency current response signal through a deep learning model; A generating module, used for inputting the estimated value of the rotor position into a motor dynamics model to generate a joint estimated value of the rotor position and speed; A control module, used for feeding back the joint estimation value to a motor control loop to perform motor vector control; The generation module is also used to determine the first rotor position and speed and the covariance matrix of the initial time step; predict the second rotor position and speed of the next time step of the initial time step according to the motor dynamics model; update the covariance matrix according to the Jacobian matrix of the motor dynamics model; calculate the Kalman gain according to the updated covariance matrix; and determine the joint estimate of the rotor position and speed according to the rotor position estimate and the Kalman gain.

8. An electronic device, characterized in that: The device comprises a memory and a processor, wherein: The processor is used to execute the computer program stored in the memory; When the processor executes the computer program, the steps of the motor vector control method based on high-frequency signal injection as described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the motor vector control method based on high-frequency signal injection as described in any one of claims 1 to 6 are implemented.

Images