A motor state estimation method, control method and system based on an improved extended Kalman filter

Abstract

This invention discloses a motor state estimation method, control method, and system based on an improved extended Kalman filter, aiming to improve the adaptability of engineering machinery under various complex working conditions. The improvement is achieved by modifying the extended Kalman filter... Q , R The matrix parameter adjustment method involves adjusting the matrix parameters in real time based on changes in the motor system parameters. This addresses the problems of complex parameter adjustment and poor angle and speed estimation performance of traditional extended Kalman filters in the face of dynamic system response (load changes). This method can effectively improve the operational reliability of construction machinery under complex working conditions, reduce maintenance costs, ensure continuous operation under extreme conditions, and further improve the dynamic response characteristics and steady-state accuracy of the entire electric drive system.

Description

Technical Field

[0001] This invention belongs to the field of electric drive motor control technology, and specifically relates to a motor state estimation method, control method and system based on an improved extended Kalman filter for electric drive systems in the field of engineering machinery. Background Technology

[0002] In electric drive systems for new energy construction machinery, high-speed permanent magnet synchronous motors (PMSMs) are widely used due to their numerous advantages. However, in existing electric drive systems, the control of PMSMs relies on the accuracy of mechanical position sensors. The installation of these sensors increases the overall size, weight, and cost of the system. Especially in the construction machinery field, harsh and continuous complex operating conditions significantly impact the accuracy and lifespan of sensors, severely reducing the reliability of construction machinery and indirectly increasing its operating costs. Therefore, sensorless motor control technology has received more widespread attention and application.

[0003] Existing sensorless motor control methods can be broadly categorized into two types: the first type is state analysis based on high-frequency signals, and the second type is state observation based on mathematical models.

[0004] The first type of state analysis method based on high-frequency signals involves superimposing a high-frequency voltage or current signal onto the fundamental signal and applying both to the three-phase windings of the motor. The corresponding high-frequency current or voltage signal carries rotor position information. By using a bandpass filter to extract this current or voltage signal and processing it appropriately, the rotor position can be estimated. Commonly used injected high-frequency signals include rotating high-frequency voltage signals and pulsating high-frequency voltage signals. This method can accurately derive the rotor position without an additional power supply, but it also has some drawbacks: the injected current leads to the generation of negative torque, and the time delay caused by the pulse injection phase limits the resolution at high speeds, resulting in a larger rotor position estimation error, thus limiting its application in high-speed scenarios.

[0005] For the second type of state observation method based on the fundamental mathematical model, these methods rely on rotational speed-related quantities (such as the generated back electromotive force) in the fundamental mathematical model to estimate the rotor position and rotational speed. Commonly used methods include sliding mode observer (SMO), model reference adaptive system (MRAS), Luenberger observer, and extended Kalman filter (EKF). These methods have good estimation performance in the medium-to-high speed range, but their estimation accuracy is poor near zero speed or extremely low speed.

[0006] In summary, the shortcomings of existing technologies are:

[0007] 1) The state observation method based on the fundamental mathematical model is valued for its excellent robustness. However, due to the inherent problem of jitter when switching sliding surfaces, the estimated rotational speed jitter and a certain phase lag are caused.

[0008] 2) The model reference adaptive method uses an adaptive control strategy to adjust parameters to match the reference model, thereby achieving accurate estimation of the motor state. However, it has problems such as sensitivity to environmental changes.

[0009] 3) Neural network algorithms are very effective in improving the accuracy and stability of control systems, but their high requirements for chip computing power limit their application in practical engineering.

[0010] 4) The Extended Kalman Filter (EKF) algorithm effectively handles system noise and measurement noise by linearizing the nonlinear system, exhibiting high robustness and estimation accuracy. However, the process noise covariance matrix Q and measurement noise covariance matrix R in the traditional EKF algorithm are usually determined by empirical trial and error, making parameter tuning cumbersome and complex. More importantly, when the motor operating conditions change, especially with sudden load changes, the fixed Q and R matrix parameters cannot be adaptively adjusted, leading to a decrease in algorithm tracking performance and an increase in speed and position estimation errors, making it difficult to meet the high reliability control requirements of complex and variable operating conditions in engineering machinery.

[0011] For example, such as Figure 6 As shown, patent CN118631122A discloses an optimization method based on a model reference adaptive system, but its core lies in the optimization of the feedback gain matrix and does not solve the EKF parameter adaptation problem. Other existing technologies, such as sliding mode observers, suffer from chattering problems, and the model reference adaptive method is highly sensitive to the environment.

[0012] Therefore, there is an urgent need for a sensorless control scheme for permanent magnet synchronous motors that can adaptively adjust key parameters and improve estimation accuracy and dynamic response performance in the full speed range, especially in the low speed range. Summary of the Invention

[0013] The purpose of this invention is to overcome the shortcomings of the prior art and provide a motor state estimation method, control method and system based on an improved extended Kalman filter. This solves the problems of the traditional extended Kalman filter, where the Q and R matrix parameters depend on manual trial and error, the adjustment process is cumbersome, and the state estimation performance degrades when the parameters are fixed due to sudden changes in motor load or speed.

[0014] In a first aspect, the present invention provides a motor state estimation method based on an improved extended Kalman filter, comprising the following steps:

[0015] S1: Acquire electrical signal data of the permanent magnet synchronous motor; the electrical signal data includes a reference voltage component and a real-time acquired current component;

[0016] S2: Based on the improved extended Kalman filter, the rotor position angle and speed of the motor are estimated in real time according to the reference voltage component and the real-time acquired current component to obtain the estimated motor state variables; the motor state variables include the rotor rotation angle and angular velocity of the motor.

[0017] The improved extended Kalman filter takes a reference voltage component and a real-time acquired current component as inputs and outputs an estimated motor state quantity. The improved extended Kalman filter adjusts the parameters of the process noise covariance matrix Q and the measurement noise covariance matrix R by introducing a nonlinear parameterized function based on the real-time motor speed and the target speed.

[0018] Furthermore, the parameter update method for the process noise covariance matrix Q includes:

[0019] Set the initial process noise covariance matrix;

[0020] The parameters of the initial process noise covariance matrix are adaptively adjusted and updated using a nonlinear parameter-determined function based on the real-time motor speed and the target speed. The nonlinear parameter-determined function of the process noise covariance matrix Q is:

[0021]

[0022] in, Let n be the diagonal elements of the initial process noise covariance matrix, and n be the estimated real-time speed of the motor. * This indicates the current target speed of the motor.

[0023] Furthermore, the parameter update method for the measurement noise covariance matrix R includes:

[0024] Set the initial measurement noise covariance matrix;

[0025] The parameters of the initial measurement noise covariance matrix are adaptively adjusted and updated using a nonlinear parameter-determined function based on the real-time motor speed and the target speed. The nonlinear parameter-determined function of the measurement noise covariance matrix R is:

[0026]

[0027] In the formula, n * This represents the current target speed of the motor, and n represents the estimated real-time speed of the motor. These are the diagonal elements of the initial measurement noise covariance matrix.

[0028] Furthermore, both the initial process noise covariance matrix and the initial measurement noise covariance matrix are set empirically.

[0029] Furthermore, based on the improved extended Kalman filter, the rotor position angle and speed of the motor are estimated in real time according to the reference voltage component and the real-time acquired current component, resulting in estimated motor state variables, including:

[0030] S21, with the current time as time k+1, predict the state vector based on the current control input. (from the α-axis reference voltage component u) αref and β-axis reference voltage component u βref (Composition) and the state estimate of k at the previous time step. Predict the state vector prediction value at time k+1. :

[0031]

[0032] Where x represents the derivative of the system's state matrix, which contains all the system's state variables. u is the system input matrix. y is the output matrix. i α i β These are the corresponding current components; ω e Let θ be the angular velocity of the motor. e The rotor rotation angle of the motor; u α u β These are the voltage components of the motor phase voltage along the α and β axes in a two-phase stationary coordinate system;

[0033] The letter k represents the state at time k, and T s Let B be the sampling period, B be the control matrix, representing the relationship between system input and output, containing all system input variables, and f() is equivalent to the system matrix;

[0034] in, This is the predicted value of the state vector at time k+1. This is the estimated state vector value of k at the previous time step. The superscript "~" represents the predicted value; the superscript "^" represents the state estimate. This is the filter input matrix at the previous time step k;

[0035] S22, predict the value based on the state vector at the current time k+1. Calculate the output corresponding to the predicted quantity ,Right now

[0036]

[0037] in, Let C be the prediction output matrix at the current time k+1, and let C be the system transfer matrix.

[0038] S23, Obtain the estimation error covariance matrix of k at the previous time step. The updated process noise covariance matrix Q and measurement noise covariance matrix R are adjusted by introducing a nonlinear parameter function based on the real-time motor speed and the target speed.

[0039] Based on the process noise covariance matrix Q and the predicted output matrix at the current time k+1 Calculate the prediction error covariance matrix :

[0040]

[0041] in, Let be the prediction error covariance matrix at the current time k+1. Let F(k) be the estimation error covariance matrix of the previous time k, with the superscript T indicating transpose; F(k) represents the state matrix of the motor system.

[0042] S24, based on the prediction error covariance matrix Calculate the gain matrix of the extended Kalman filter based on the measured noise covariance matrix R. :

[0043]

[0044] Where K(k+1) is the gain matrix of the extended Kalman filter at the current time k+1, C is the transfer matrix of the system, and the superscript T in the right indicates transpose;

[0045] S25, predict the state vector at time k+1 based on the gain matrix K(k+1) of the extended Kalman filter. Perform feedback correction to obtain the state estimate at time k+1. :

[0046]

[0047] in, This is the state estimate at time k+1, i.e., the filter output.

[0048] Furthermore, the method also includes:

[0049] S26, Calculate the estimation error covariance matrix of the current time k+1 for the next time step estimation in advance according to the following formula. :

[0050]

[0051] in, Let be the estimation error covariance matrix at the current time k+1. Let K(k+1) be the prediction error covariance matrix at the current time k+1, K(k+1) be the gain matrix of the extended Kalman filter at the current time k+1, and C be the transfer matrix of the system.

[0052] The estimation error covariance matrix at the current time k+1 is calculated. Then, proceed to the next moment's motor state estimation.

[0053] Secondly, the present invention provides a sensorless control method for a permanent magnet synchronous motor based on an improved extended Kalman filter, comprising:

[0054] Acquire electrical signal data of a permanent magnet synchronous motor; the electrical signal data includes a reference voltage component and a real-time acquired current component;

[0055] Based on the motor state estimation method based on the improved extended Kalman filter as described in the first aspect, the estimated motor state variables are obtained; the motor state variables include the rotor rotation angle and angular velocity of the motor.

[0056] Using the estimated motor state variables, the motor is subjected to dual closed-loop vector control of speed and torque, and a space vector pulse width modulation signal is generated to drive the three-phase full-bridge inverter connected to the permanent magnet synchronous motor.

[0057] Furthermore, using the estimated motor state variables, dual closed-loop vector control of speed and torque is performed on the motor, and a space vector pulse width modulation signal is generated to drive a three-phase full-bridge inverter connected to the permanent magnet synchronous motor, including:

[0058] The target motor speed is obtained based on the motor speed command;

[0059] Convert the motor's real-time angular velocity into real-time rotational speed;

[0060] The difference between the target motor speed and the real-time speed is fed into the speed loop PI controller to obtain the reference values ​​of the d-axis and q-axis currents.

[0061] The real-time values ​​of the d-axis and q-axis currents are compared with the reference values ​​of the d-axis and q-axis currents, respectively, and the difference is input to the torque loop PI controller to output the d-axis and q-axis reference voltages. The real-time values ​​of the d-axis and q-axis currents are obtained by converting the three-phase currents output to the permanent magnet synchronous motor by the three-phase inverter through Clark transformation and then undergoing coordinate transformation.

[0062] After the d-axis and q-axis reference voltages are converted into α-axis and β-axis reference voltages through coordinate conversion, they are uniformly modulated by SVPWM to obtain the switching sequence control signals of the 6 IGBTs in the three-phase full-bridge inverter, so as to drive the three-phase full-bridge inverter connected to the permanent magnet synchronous motor.

[0063] Furthermore, the reference voltage component and the real-time acquired current component include the α-axis reference voltage component u. αref β-axis reference voltage component u βref α-axis current component i α β-axis current component i β ;

[0064] Furthermore, obtain the α-axis current component i α β-axis current component i β The methods include:

[0065] Obtain the three-phase current output from the three-phase full-bridge inverter to the permanent magnet synchronous motor;

[0066] Perform Clark transform on the three-phase currents to obtain the α-axis current component i. α β-axis current component i β .

[0067] Furthermore, methods for obtaining the reference voltage component include:

[0068] Obtain the d-axis and q-axis reference voltages output by the torque loop PI controller;

[0069] Based on the d-axis and q-axis reference voltages, the coordinate transformation is used to obtain the α-axis and β-axis reference voltage components u. αref and β-axis reference voltage component u βref .

[0070] Thirdly, the present invention provides a sensorless control system for a permanent magnet synchronous motor based on an improved extended Kalman filter, comprising: a permanent magnet synchronous motor, a three-phase full-bridge inverter, a current sampling module, and a controller;

[0071] The current sampling module is connected to the permanent magnet synchronous motor and is used to collect the three-phase current output from the three-phase full-bridge inverter to the permanent magnet synchronous motor, and convert it into the α-axis current component i. α β-axis current component i β Send to the controller;

[0072] The controller is connected to the three-phase full-bridge inverter and the current sampling module respectively, and is used to execute the control method as described in the second aspect, generate a space vector pulse width modulation signal, and output it to the three-phase full-bridge inverter, thereby driving the three-phase full-bridge inverter connected to the permanent magnet synchronous motor to operate.

[0073] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0074] 1. The improved Extended Kalman Filter (EKF) of this invention significantly enhances the system's adaptive capability by introducing a nonlinear parameter-fixed function to dynamically adjust the parameters of matrices Q and R. When the motor's operating state changes, the system can calculate the optimal Q and R matrix parameter values ​​based on real-time data, avoiding the increased estimation error caused by fixed parameters in traditional methods. This adaptive adjustment mechanism not only improves the estimation accuracy of rotor position and speed but also enhances the system's robustness under complex operating conditions. By adaptively adjusting the Q and R matrices through a nonlinear parameter-fixed function, the tedious manual parameter trial-and-error process in traditional EKF is completely avoided, and the filter parameters can be automatically optimized based on the motor's real-time operating state (speed error), improving the algorithm's adaptive capability and ease of engineering application.

[0075] 2. To further optimize algorithm performance, this invention comprehensively considers the dynamic characteristics of the motor and the noise distribution pattern in the design of the nonlinear parameter function. Through analysis and modeling of historical data, the optimal parameter range suitable for different operating conditions is determined, and fine-tuning is performed using real-time feedback information to ensure that the algorithm maintains high-precision estimation across the entire speed domain. Simultaneously, this method effectively suppresses the chattering phenomenon commonly seen in traditional extended Kalman filters in the low-speed region, enabling the system to maintain good stability even at zero or extremely low speeds.

[0076] 3. This invention can significantly reduce the reliance of construction machinery on mechanical sensors, thereby reducing maintenance costs and downtime caused by sensor failures. Especially under extreme conditions, such as high-load shocks or frequent start-stop scenarios, this technology can significantly improve the reliability and continuous operation capability of the electric drive system, providing strong technical support for the intelligent development of new energy construction machinery. By improving the accuracy and robustness of control, this solution significantly enhances the operational reliability and continuity of the electric drive system of construction machinery under complex and harsh conditions, reducing the downtime risk and maintenance costs caused by sensor failures.

[0077] 4. Improved dynamic and steady-state performance: The adaptive mechanism enables the EKF to quickly adjust its gain during dynamic processes such as load changes and speed variations, improving the system's tracking speed and accuracy of rotor position and speed, and reducing overshoot and steady-state errors.

[0078] 5. Improved low-speed performance: Targeted parameter adjustment strategies effectively improve the observation accuracy of traditional sensorless control methods based on fundamental wave models at low speeds and near zero speeds, and broaden the speed range for stable sensorless operation. Attached Figure Description

[0079] Figure 1This is a block diagram of the filtering estimation principle based on the improved EKF in an embodiment of the present invention.

[0080] Figure 2 This is an overall control structure diagram of the sensorless control system for permanent magnet synchronous motor in an embodiment of the present invention.

[0081] Figure 3 This is a flowchart of a sensorless control method based on an improved extended Kalman filter in an embodiment of the present invention.

[0082] Figure 4 This is a design flowchart of the improved extended Kalman filter in an embodiment of the present invention.

[0083] Figure 5 This is a schematic diagram comparing the motor speed and angle response curves of the embodiment of the present invention with those of the traditional EKF algorithm.

[0084] Figure 6 It is the control structure diagram of patent CN118631122A. Detailed Implementation

[0085] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0086] Example 1:

[0087] One embodiment of the present invention provides a motor state estimation method based on an improved extended Kalman filter, the prediction principle of which is as follows: Figure 1 As shown.

[0088] Appendix Figure 1 The present invention describes the filtering estimation principle structure based on the improved extended Kalman filter in the implementation example. The controlled object compares the system estimation model with the system estimation model by controlling the input, based on the system noise and the observation noise, and finally adjusts the estimation model in real time by the improved extended Kalman filter gain.

[0089] like Figure 2 As shown, the system that this method relies on includes a permanent magnet synchronous motor, a three-phase full-bridge inverter powered by a DC power supply, a current sampling module (not shown separately in the figure, used to collect the three-phase current of the motor), and a controller (usually a DSP or MCU).

[0090] This system eliminates the need for a rotary transformer structure in a sensorless control system. In this application, the motor state quantities refer to rotor position information, specifically the rotation angle and angular velocity of the motor rotor. For simplicity, these are referred to as the motor's rotation angle and angular velocity.

[0091] like Figure 1 and Figure 3As shown, its control flow mainly includes:

[0092] S1: Establish and discretize the mathematical model of the motor. Specifically, in the two-phase stationary α-β coordinate system, establish a model with current i... α i β With angular velocity ω and motor angle as state variables x, and voltage u as the state variable... α ,u β The input state-space equation is discretized to obtain the form as follows:

[0093]

[0094] Discrete model.

[0095] S2: Design an improved Extended Kalman Filter (EKF) for state estimation. This is the core improvement of this invention, and its detailed process is as follows: Figure 4 As shown, the improved Extended Kalman Filter (EKF) of this invention significantly enhances the system's adaptive capability by introducing a nonlinear parameter-fixed function to dynamically adjust the parameters of matrices Q and R. Specifically, when the motor's operating state changes, such as sudden load changes or speed fluctuations, the system can calculate the optimal Q and R matrix parameter values ​​based on real-time data, avoiding the increased estimation error caused by fixed parameters in traditional methods. This adaptive adjustment mechanism not only improves the estimation accuracy of rotor position and speed but also enhances the system's robustness under complex operating conditions. By adaptively adjusting the Q and R matrices through a nonlinear parameter-fixed function, the tedious manual parameter trial-and-error process in traditional EKF is completely avoided, and the filter parameters can be automatically optimized based on the motor's real-time operating state (speed error), improving the algorithm's adaptive capability and ease of engineering application.

[0096] The method specifically includes:

[0097] S21, with the current time as time k+1, predict the state vector based on the current control input. The state estimate of k at the previous time step Predict the state vector prediction value at time k+1. :

[0098]

[0099] Where x represents the derivative of the system's state matrix, which contains all the system's state variables. u is the system input matrix. ,at this time The α-axis reference voltage component u from the previous controller output is acquired. αref and β-axis reference voltage component u βref Composition, where y is the output matrix, i α i β These are the corresponding current components; ω e Let θ be the angular velocity of the motor. e The rotor rotation angle of the motor; u α u β These are the voltage components of the motor phase voltage along the α and β axes in a two-phase stationary coordinate system;

[0100] The letter k represents the state at time k, and T s Let B be the sampling period, B be the control matrix, representing the relationship between system input and output, containing all system input variables, and f() is equivalent to the system matrix;

[0101] in, This is the predicted value of the state vector at time k+1. This is the estimated state vector value of k at the previous time step. The superscript "~" represents the predicted value; the superscript "^" represents the state estimate. This is the filter input matrix at the previous time step k;

[0102] S22, predict the value based on the state vector at the current time k+1. Calculate the output corresponding to the predicted quantity ,Right now

[0103]

[0104] in, Let C be the prediction output matrix at the current time k+1, and let C be the system transfer matrix.

[0105] S23, Obtain the estimation error covariance matrix of k at the previous time step. The updated process noise covariance matrix Q and measurement noise covariance matrix R are adjusted by introducing a nonlinear parameter function based on the real-time motor speed and the target speed.

[0106] Methods for updating the parameters of the process noise covariance matrix Q include:

[0107] Set the initial process noise covariance matrix;

[0108] The parameters of the initial process noise covariance matrix are adaptively adjusted and updated using a nonlinear parameter-determined function based on the real-time motor speed and the target speed. The nonlinear parameter-determined function of the process noise covariance matrix Q is:

[0109]

[0110] in, Let n be the diagonal elements of the initial process noise covariance matrix, and n be the estimated real-time speed of the motor. *This indicates the current target speed of the motor.

[0111] Methods for updating the parameters of the noise covariance matrix R include:

[0112] Set the initial measurement noise covariance matrix;

[0113] The parameters of the initial measurement noise covariance matrix are adaptively adjusted and updated using a nonlinear parameter-determined function based on the real-time motor speed and the target speed. The nonlinear parameter-determined function of the measurement noise covariance matrix R is:

[0114]

[0115] In the formula, n * This represents the current target speed of the motor, and n represents the estimated real-time speed of the motor. These are the diagonal elements of the initial measurement noise covariance matrix.

[0116] Preferably, the initial process noise covariance matrix and the initial measurement noise covariance matrix are both set empirically.

[0117] Based on the process noise covariance matrix Q and the predicted output matrix at the current time k+1 Calculate the prediction error covariance matrix :

[0118]

[0119] in, Let be the prediction error covariance matrix at the current time k+1. Let F(k) be the estimation error covariance matrix of the previous time k, with the superscript T indicating transpose; F(k) represents the state matrix of the motor system.

[0120] in:

[0121]

[0122] The result is

[0123]

[0124] F(k) represents the state of the motor system at time k, which includes the current values ​​of each motor parameter and state variable.

[0125] S24, based on the prediction error covariance matrix Calculate the gain matrix of the extended Kalman filter based on the measured noise covariance matrix R. :

[0126]

[0127] Where K(k+1) is the gain matrix of the extended Kalman filter at the current time k+1, C is the transfer matrix of the system, and the superscript T in the right indicates transpose;

[0128] S25, predict the state vector at time k+1 based on the gain matrix K(k+1) of the extended Kalman filter. Perform feedback correction to obtain the state estimate at time k+1. :

[0129]

[0130] in, This is the state estimate at time k+1, i.e., the filter output.

[0131] S26, Calculate the estimation error covariance matrix of the current time k+1 for the next time step estimation in advance according to the following formula. :

[0132]

[0133] in, Let be the estimation error covariance matrix at the current time k+1. Let K(k+1) be the prediction error covariance matrix at the current time k+1, K(k+1) be the gain matrix of the extended Kalman filter at the current time k+1, and C be the transfer matrix of the system.

[0134] Appendix Figure 5 The figure shows the response curves of motor speed and angle in the implementation examples of this invention, compared with those of the traditional extended Kalman filter. It can be seen from the figure that under the control of the sensorless control system and method of permanent magnet synchronous motor based on the technical solution of this invention: the steady-state estimation accuracy and dynamic tracking accuracy of motor position and speed can be greatly improved.

[0135] 1) Improve the accuracy of motor position (angle) and steady-state estimation and dynamic tracking under complex working conditions with sudden load changes in the existing sensorless control strategy for permanent magnet synchronous motors;

[0136] 2) Improve the control performance and design efficiency of traditional sensorless motor control algorithms and controllers based on extended Kalman filtering;

[0137] 3) Improve the estimation accuracy of conventional sensorless motor control methods based on fundamental wave models near low speed or zero speed.

[0138] In the above embodiments, the adaptive function of the Q and R matrices is the core, but its specific function form and nonlinear relationship (such as exponential, linear combination, etc.) can be adjusted and optimized according to different motor models and application scenarios. These transformations and selections all fall within the protection scope of this invention.

[0139] Example 2:

[0140] This embodiment provides a sensorless motor control method based on an improved extended Kalman filter, including but not limited to the following steps:

[0141] S1. Establish the mathematical model of the permanent magnet synchronous motor. Combining Kirchhoff's voltage and current laws and the Park coordinate transformation principle, the voltage and current state equations of the permanent magnet synchronous motor in the two-phase stationary coordinate system are summarized as follows:

[0142] ;

[0143] ;

[0144] Among them, u α u β These are the voltage components of the motor phase voltage along the αβ axis in a two-phase stationary coordinate system. Similarly, i α i β These are the corresponding current components; R and Ls are the stator resistance and stator inductance of the motor, respectively; ω e ѱ represents the angular velocity of the motor. f For permanent magnet flux linkage, θ e d / dt represents the rotor rotation angle of the motor; d / dt is the differential symbol, and sin / cos is the sine / cosine function.

[0145] S2, combining the state-space expression from linear system theory, the mathematical model of the motor is rearranged into the corresponding form, and the motor mathematical model is discretized to facilitate the subsequent design of the controller and the observation of the extended Kalman filter. The basic form of the state-space expression is as follows:

[0146]

[0147] In the formula, x represents the derivative of the system's state matrix, containing all the system's state variables; A is the system matrix, representing some inherent parameters of the system; B is the control matrix, representing the relationship between the system's input and output; u is the system's input matrix, containing all the system's input variables; C is the system's transfer matrix, representing the transfer relationship between the system's input and output; and y is the output matrix, containing all the system's output variables. Furthermore, f(x) is equivalent to Ax.

[0148] Combining the motor model in S1, where , , , , , ;

[0149] Furthermore, the discrete mathematical model of the motor:

[0150]

[0151] Where W(k) and V(k) are the measurement noise and system noise, respectively; the letter k represents the state at time k, and W(k) and V(k) are the measurement noise and system noise, respectively.

[0152] S3. Design an improved extended Kalman filter to estimate the motor position (angle) and speed, and define the covariance matrices Q and R as follows:

[0153]

[0154] Where Q and R represent the covariance matrices of the noise matrices of measurement noise and system noise, respectively. In general, p: error covariance matrix, Q: (system noise) covariance matrix, R: (measurement noise) covariance matrix.

[0155] S4. Based on the improved extended Kalman filter, the estimated real-time motor position (angle) and speed are obtained, and a dual closed-loop control strategy for speed and torque is designed for the motor.

[0156] S5, Design a PI controller for the speed and torque dual closed-loop control strategy in step S4;

[0157] S6. Design a space vector modulation algorithm for the speed and torque dual closed-loop control strategy in step S4 to obtain the final switching sequence control signal of the three-phase full-bridge inverter IGBT.

[0158] Based on the improved extended Kalman filter in step S3 of the above technical solution, combined with Figure 2 It can be seen more clearly that

[0159] The improved extended Kalman filter input is the α-axis reference voltage component U. αref β-axis reference voltage component U βref α-axis current component i α β-axis current component i β The output is the observed rotor rotation angle θ of the motor. e .

[0160] like Figure 4 As shown, the improved extended Kalman filter execution method includes the following steps:

[0161] S31, predict the state vector, i.e., based on the control input. Compared to the previous state estimate To predict the state vector at time (k+1), it should be:

[0162]

[0163] Where Ts is the sampling period, "^" represents the state estimate, and "~" represents the predicted value.

[0164] S32, Calculate the output estimate at time K+1 corresponding to the prediction in step S1 of the improved extended Kalman filter design. ,Right now

[0165]

[0166] in, Let y represent the output estimate at time k+1, where y represents the system output.

[0167] S33, Calculate the error covariance matrix That is:

[0168]

[0169] Where p represents the error matrix of the covariance matrix Q, which is also the prediction mean square error matrix.

[0170] Where F represents the state of the motor system, including the current values ​​of various motor parameters and state variables:

[0171]

[0172] The result is

[0173]

[0174] F(k) represents the state of the motor system at time k, which includes the current values ​​of each motor parameter and state variable.

[0175] In particular, in traditional extended Kalman filters, the parameter optimization of the covariance matrix Q is generally done by trial and error based on engineering experience. This not only makes the overall design of the extended Kalman filter cumbersome and inefficient, but also greatly affects the estimation accuracy of motor position (angle) and the tracking accuracy of motor speed.

[0176] To overcome the shortcomings of traditional extended Kalman filter designs and further improve the estimation accuracy of the motor near zero speed, this technical solution first sets an initial process noise covariance matrix based on experience, and then adaptively adjusts and updates it according to a nonlinear parameterized function based on the real-time motor speed and the target speed. Specifically, this includes:

[0177] First, the initial process noise covariance matrix is ​​set based on experience.

[0178] Then, the parameters of the initial process noise covariance matrix are adaptively adjusted and updated using a nonlinear parameter-determined function based on the real-time motor speed and the target speed. Specifically, the nonlinear parameter-determined function of the process noise covariance matrix Q is:

[0179]

[0180] Where q1, q2, q3, q4 are the diagonal elements of the initial process noise covariance matrix, and n is the estimated real-time speed of the motor. * This indicates the current target speed of the motor.

[0181] S34, Calculate the gain matrix of the extended Kalman filter. That is,

[0182]

[0183] In particular, traditional extended Kalman filters typically use a trial-and-error method to optimize the parameters of the covariance matrix R, requiring repeated adjustments through experimental testing. This significantly impacts the convergence and fluctuation suppression of motor speed. To improve the efficiency of parameter trial-and-error and the steady-state accuracy and dynamic convergence speed of motor speed control, this technical solution adopts the following nonlinear time-varying covariance matrix R parameter optimization method, the specific selection method of which is as follows:

[0184] The parameters of the noise covariance matrix R are first set empirically, and then adaptively adjusted and updated using a nonlinear parameter-determined function based on the real-time motor speed and the target speed. Specifically, this includes:

[0185] First, an initial measurement noise covariance matrix is ​​set based on experience.

[0186] Then, the parameters of the initial measurement noise covariance matrix are adaptively adjusted and updated using a nonlinear parameter-determined function based on the real-time motor speed and the target speed. Specifically, the nonlinear parameter-determined function for the measurement noise covariance matrix R is as follows:

[0187]

[0188] In the formula, n *The target speed of the motor is represented by , n represents the estimated real-time speed of the motor, and r1 and r2 are the diagonal elements of the initial measurement noise covariance matrix.

[0189] S35, for the predicted state vector Feedback correction is performed to obtain an optimized state estimate. ,Right now

[0190]

[0191] in, This is the state estimate at time k+1, i.e., the filter output.

[0192] This step is also called corrected state estimation, or "filtering".

[0193] S36, For the next estimation, the estimation error covariance matrix needs to be calculated in advance, i.e., we have

[0194]

[0195] Where p is the error covariance matrix, Q is the (system noise) covariance matrix, and R is the (measurement noise) covariance matrix. Let be the estimation error covariance matrix at the current time k+1. Let K(k+1) be the prediction error covariance matrix at the current time k+1, K(k+1) be the gain matrix of the extended Kalman filter at the current time k+1, and C be the transfer matrix of the system.

[0196] In summary, based on the improved extended Kalman filter sensorless control method for permanent magnet synchronous motors proposed in this invention, the motor can operate well within a wide speed range, and in particular, it can significantly improve the control effect of traditional extended Kalman filters near the zero speed of the motor.

[0197] In this embodiment, after evaluating and obtaining the estimated real-time motor state, including the position (angle) and speed of the motor rotor, the next step of control is to be performed based on the real-time motor state:

[0198] S5, Design a PI controller for the speed and torque dual closed-loop control strategy in step S4, including:

[0199] Using the estimated motor state variables, the motor is subjected to dual closed-loop vector control of speed and torque, and a space vector pulse width modulation signal is generated to drive the three-phase full-bridge inverter connected to the permanent magnet synchronous motor.

[0200] Specifically, dual closed-loop vector control is implemented. The estimated rotational speed is compared with the given rotational speed, and the q-axis current reference value is output through the speed PI controller; the estimated d-axis and q-axis currents (obtained through Clark transformation and Park transformation) are compared with their respective reference values, and the d-axis and q-axis reference voltages are output through the current (torque) PI controller.

[0201] S6, A space vector modulation algorithm is designed for the speed-torque dual closed-loop control strategy in step S4 to obtain the final switching sequence control signals for the three-phase full-bridge inverter IGBTs, including:

[0202] The drive signal is generated. The d-axis and q-axis reference voltages are inversely Park-transformed and then modulated by the SVPWM module to generate the switching signals for driving the IGBTs in the three-phase full-bridge inverter, thereby controlling the motor operation.

[0203] The specific control process is as follows: Figure 2 As shown, it includes:

[0204] The target motor speed is obtained based on the motor speed command;

[0205] Convert the motor's real-time angular velocity into real-time rotational speed;

[0206] The difference between the target motor speed and the real-time speed is fed into the speed loop PI controller to obtain the reference values ​​of the d-axis and q-axis currents.

[0207] The real-time values ​​of the d-axis and q-axis currents are compared with the reference values ​​of the d-axis and q-axis currents, respectively, and the difference is input to the torque loop PI controller to output the d-axis and q-axis reference voltages. The real-time values ​​of the d-axis and q-axis currents are obtained by converting the three-phase currents output to the permanent magnet synchronous motor by the three-phase inverter through Clark transformation and then undergoing coordinate transformation.

[0208] After the d-axis and q-axis reference voltages are converted into α-axis and β-axis reference voltages through coordinate conversion, they are uniformly modulated by SVPWM to obtain the switching sequence control signals of the 6 IGBTs in the three-phase full-bridge inverter, so as to drive the three-phase full-bridge inverter connected to the permanent magnet synchronous motor.

[0209] The reference voltage components and the real-time acquired current components include the α-axis reference voltage component u. αref β-axis reference voltage component u βref α-axis current component i α β-axis current component i β ;

[0210] Specifically, obtain the α-axis current component i α β-axis current component i β The methods include:

[0211] Obtain the three-phase current output from the three-phase full-bridge inverter to the permanent magnet synchronous motor;

[0212] Perform Clark transform on the three-phase currents to obtain the α-axis current component i. α β-axis current component i β .

[0213] Specifically, methods for obtaining the reference voltage component include:

[0214] Obtain the d-axis and q-axis reference voltages output by the torque loop PI controller (which is the control input of the speed and torque dual closed-loop PI controller).

[0215] Based on the d-axis and q-axis reference voltages, the coordinate transformation is used to obtain the α-axis and β-axis reference voltage components u. αref and β-axis reference voltage component u βref .

[0216] Specifically, attached Figure 2 To implement the invention, the permanent magnet synchronous motor system based on the improved extended Kalman filter is described. For sensorless control of the permanent magnet synchronous motor, the primary components are a three-phase full-bridge inverter and a DC power supply. The specific control system implementation is as follows: First, the motor speed command is set. The difference between this command and the real-time speed is compared to the q-axis reference current output by the PI controller, resulting in a q-axis reference current. This q-axis reference current is then compared to the real-time q-axis reference current, again resulting in a q-axis reference voltage output by the PI controller. Simultaneously, the d-axis reference current is compared to the real-time d-axis reference current, again resulting in a d-axis reference voltage output by the PI controller. The real-time d and q-axis currents are obtained through a Clar converter module and then a Park converter module, all three-phase currents. Finally, these are uniformly modulated by the SVPWM module to obtain the switching sequence control signals for the six IGBTs in the three-phase full-bridge inverter. SVPWM is short for Space Vector Pulse Width Modulation.

[0217] Figure 5 The simulation results comparing the proposed solution with the traditional fixed-parameter EKF solution under the same operating conditions are presented. It can be seen that when the load or speed changes abruptly, the proposed solution provides a faster speed estimation response, smaller overshoot, and higher steady-state accuracy, verifying its superiority.

[0218] Example 3:

[0219] This embodiment provides a sensorless control system for a permanent magnet synchronous motor based on an improved extended Kalman filter, including: a permanent magnet synchronous motor, a three-phase full-bridge inverter, a current sampling module, and a controller;

[0220] The current sampling module is connected to the permanent magnet synchronous motor and is used to collect the three-phase current output from the three-phase full-bridge inverter to the permanent magnet synchronous motor, and convert it into the α-axis current component i. αβ-axis current component i β Send to the controller;

[0221] The controller is connected to the three-phase full-bridge inverter and the current sampling module respectively, and is used to execute the control method as described in Example 2, generate a space vector pulse width modulation signal, and output it to the three-phase full-bridge inverter, thereby driving the three-phase full-bridge inverter connected to the permanent magnet synchronous motor to operate.

[0222] Specifically, attached Figure 2 To implement the invention, the permanent magnet synchronous motor system based on the improved extended Kalman filter system control structure is described. For the sensorless control of the permanent magnet synchronous motor, the main components are a three-phase full-bridge inverter and a DC power supply. The specific control system is implemented as follows: First, the motor speed command is set. The q-axis reference current is obtained by comparing the difference with the real-time speed and outputting it through the PI controller. The q-axis reference voltage is obtained by comparing it with the real-time q-axis current and outputting it through the PI controller. Simultaneously, the d-axis reference current is compared with the real-time d-axis reference current and outputting it through the PI controller to obtain the d-axis reference voltage. The real-time d and q-axis currents are obtained by the three-phase current through the Clar conversion module and then through the Park conversion module. Finally, they are uniformly modulated by the SVPWM module to obtain the switching sequence control signals of the six IGBTs in the three-phase full-bridge inverter.

[0223] The instructions executed by the controller are as follows: Figure 3-4 As shown:

[0224] Appendix Figure 3 For the implementation use case of the invention, a sensorless control method for permanent magnet synchronous motors based on an improved extended Kalman filter is described:

[0225] S1. Establish a mathematical model of the permanent magnet synchronous motor and obtain the voltage and current equations in the two-phase stationary coordinate system through coordinate transformation.

[0226] S2, combining the state-space expression in linear system theory, discretizes the mathematical model of the motor;

[0227] S3, design an improved extended Kalman filter to estimate the motor position (angle) and speed, and define the covariance matrices Q and R;

[0228] S4. Based on the improved extended Kalman filter, the estimated real-time motor position (angle) and speed are obtained, and a speed and torque closed-loop control strategy for the motor is designed.

[0229] S5, Design of a PI controller for the dual closed-loop control strategy of speed and torque;

[0230] S6. A space vector modulation algorithm is designed for the dual closed-loop control strategy of speed and torque to obtain the final switching sequence control signal of the IGBT of the three-phase full-bridge inverter.

[0231] Appendix Figure 4 The prediction process of the improved Kalman filter in the implementation examples of this invention is as follows:

[0232] S1, predict the state vector, that is, predict the state vector at time (k+1) based on the input u(k) and the previous state estimate;

[0233] S2, Calculate the output corresponding to the predicted quantity in step S1 of the improved extended Kalman filter design;

[0234] S3, calculate the error covariance matrix Q.

[0235] In particular, to further improve the estimation accuracy of the motor near zero speed, this technical solution adopts a nonlinear time-varying covariance matrix Q parameter optimization method;

[0236] S4, Calculate the gain matrix K(k+1) of the improved extended Kalman filter.

[0237] In particular, to improve the efficiency of parameter trial and error and the steady-state accuracy and dynamic convergence speed of motor speed control, this technical solution adopts the following nonlinear time-varying covariance matrix R parameter optimization method;

[0238] S5, perform feedback correction on the predicted state vector to obtain an optimized state estimate;

[0239] S6 is for the next estimation, where the estimation error covariance matrix is ​​pre-calculated.

[0240] Figure 5 The simulation results comparing the proposed solution with the traditional fixed-parameter EKF solution under the same operating conditions are presented. It can be seen that when the load or speed changes abruptly, the proposed solution provides a faster speed estimation response, smaller overshoot, and higher steady-state accuracy, verifying its superiority.

[0241] There are other alternative solutions to the technical solution of the present invention. For example, by further optimizing and improving step S3 of the sensorless control method for permanent magnet synchronous motors based on improved extended Kalman filtering, as well as steps S3 and S4 of the improved extended Kalman filter design, such as by designing nonlinear value functions for the covariance matrices Q and R to adaptively improve the technical solution of the present invention for the controlled object, it is possible to reproduce the technical solution of the present invention and achieve the same purpose.

[0242] This invention can significantly reduce the reliance of construction machinery on mechanical sensors, thereby reducing maintenance costs and downtime caused by sensor failures. Especially under extreme conditions, such as high-load shocks or frequent start-stop scenarios, this technology can significantly improve the reliability and continuous operation capability of the electric drive system, providing strong technical support for the intelligent development of new energy construction machinery. By improving the precision and robustness of control, this solution significantly enhances the operational reliability and continuity of the electric drive system of construction machinery under complex and harsh conditions, reducing the downtime risk and maintenance costs caused by sensor failures.

[0243] This invention can significantly improve dynamic and steady-state performance: the adaptive mechanism enables the EKF to quickly adjust the gain during dynamic processes such as load changes and speed variations, thereby improving the system's tracking speed and accuracy of rotor position and speed, and reducing overshoot and steady-state errors.

[0244] This invention can improve low-speed performance: the targeted parameter adjustment strategy effectively improves the observation accuracy of traditional sensorless control methods based on fundamental wave models at low speeds and near zero speeds, and broadens the speed range for stable sensorless operation.

[0245] This invention can also be used in sensorless motor control methods in other industrial variable frequency speed control fields related to automotive electronics and power electronics.

[0246] This embodiment also protects a computer-readable storage medium (such as ROM, RAM, disk, optical disk, etc.) that stores a computer program. When the program is executed by a processor (such as a DSP of a motor controller), it can implement the process steps of the above method embodiment.

[0247] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0248] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0249] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0250] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0251] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for estimating motor state based on an improved extended Kalman filter, characterized in that, Includes the following steps: S1: Acquire electrical signal data of the permanent magnet synchronous motor; the electrical signal data includes a reference voltage component and a real-time acquired current component; S2: Based on the improved extended Kalman filter, the rotor position angle and speed of the motor are estimated in real time according to the reference voltage component and the real-time acquired current component to obtain the estimated motor state variables; the motor state variables include the rotor rotation angle and angular velocity of the motor. The improved extended Kalman filter takes a reference voltage component and a real-time acquired current component as inputs and outputs an estimated motor state quantity. The improved extended Kalman filter adjusts the parameters of the update process noise covariance matrix Q and the measurement noise covariance matrix R by introducing a nonlinear parameterized function based on the real-time motor speed and the target speed. Based on the improved extended Kalman filter, the rotor position angle and speed of the motor are estimated in real time according to the reference voltage component and the real-time acquired current component, resulting in estimated motor state variables, including: S21, predict the state vector using the filter input matrix from the previous time step k. The state estimate of k at the previous time step Predict the state vector prediction value at time k+1. : ； Where x represents the state variable, u is the filter input matrix. y is the output matrix. i α i β ω represents the current components along the α and β axes, respectively; e Let θ be the angular velocity of the motor. e The rotor rotation angle of the motor; u α u β These are the voltage components of the motor phase voltage along the α and β axes in a two-phase stationary coordinate system; The letter k represents the state at time k, and T s Let B be the sampling period, B be the control matrix, representing the relationship between the system input and output, and f() be equivalent to the system matrix; in, This is the predicted value of the state vector at time k+1. This is the estimated state vector value of k at the previous time step. The superscript "~" represents the predicted value; the superscript "^" represents the state estimate. This is the filter input matrix at the previous time step k; S22, predict the value based on the state vector at the current time k+1. Calculate the output corresponding to the predicted quantity ,Right now ； in, Let C be the prediction output matrix at the current time k+1, and let C be the system transfer matrix. S23, Obtain the estimation error covariance matrix of k at the previous time step. The updated process noise covariance matrix Q and measurement noise covariance matrix R are adjusted by introducing a nonlinear parameter function based on the real-time motor speed and the target speed. Based on the process noise covariance matrix Q and the predicted output matrix at the current time k+1 Calculate the prediction error covariance matrix : ； in, Let be the prediction error covariance matrix at the current time k+1. Let F(k) be the estimation error covariance matrix of the previous time k, with the superscript T indicating transpose; F(k) represents the state matrix of the motor system. S24, based on the prediction error covariance matrix Calculate the gain matrix of the extended Kalman filter based on the measured noise covariance matrix R. : ； Where K(k+1) is the gain matrix of the extended Kalman filter at the current time k+1, C is the transfer matrix of the system, and the superscript T in the right indicates transpose; S25, predict the state vector at time k+1 based on the gain matrix K(k+1) of the extended Kalman filter. Perform feedback correction to obtain the state estimate at time k+1. : ； in, y(k+1) is the state estimate at time k+1, i.e., the filter output; y(k+1) is the system output matrix at time k+1.

2. The motor state estimation method based on the improved extended Kalman filter according to claim 1, characterized in that, The parameter update method for the process noise covariance matrix Q includes: Set the initial process noise covariance matrix; The parameters of the initial process noise covariance matrix are adaptively adjusted and updated using a nonlinear parameter-determined function based on the real-time motor speed and the target speed. The nonlinear parameter-determined function of the process noise covariance matrix Q is: ； in, Let n be the diagonal elements of the initial process noise covariance matrix, and n be the estimated real-time speed of the motor. * This indicates the current target speed of the motor.

3. The motor state estimation method based on the improved extended Kalman filter according to claim 2, characterized in that, The parameter update method for the measurement noise covariance matrix R includes: Set the initial measurement noise covariance matrix; The parameters of the initial measurement noise covariance matrix are adaptively adjusted and updated using a nonlinear parameter-determined function based on the real-time motor speed and the target speed. The nonlinear parameter-determined function of the measurement noise covariance matrix R is: ； In the formula, n * This represents the current target speed of the motor, and n represents the estimated real-time speed of the motor. These are the diagonal elements of the initial measurement noise covariance matrix.

4. The motor state estimation method based on the improved extended Kalman filter according to claim 3, characterized in that, The initial process noise covariance matrix and the initial measurement noise covariance matrix are both set empirically.

5. The motor state estimation method based on the improved extended Kalman filter according to claim 4, characterized in that, The method further includes: S26, Calculate the estimation error covariance matrix of the current time k+1 for the next time step estimation in advance according to the following formula. : ； in, Let be the estimation error covariance matrix at the current time k+1. Let K(k+1) be the prediction error covariance matrix at the current time k+1, K(k+1) be the gain matrix of the extended Kalman filter at the current time k+1, and C be the transfer matrix of the system. The estimation error covariance matrix at the current time k+1 is calculated. Then, proceed to the next moment's motor state estimation.

6. A sensorless control method for a permanent magnet synchronous motor based on an improved extended Kalman filter, characterized in that, include: Acquire electrical signal data of permanent magnet synchronous motor; The electrical signal data includes a reference voltage component and a real-time acquired current component. Based on the motor state estimation method based on the improved extended Kalman filter as described in any one of claims 1-5, the estimated motor state variables are obtained; the motor state variables include the rotor rotation angle and angular velocity of the motor. Using the estimated motor state variables, the motor is subjected to dual closed-loop vector control of speed and torque, and a space vector pulse width modulation signal is generated to drive the three-phase full-bridge inverter connected to the permanent magnet synchronous motor.

7. The sensorless control method for permanent magnet synchronous motors based on improved extended Kalman filtering according to claim 6, characterized in that, Using the estimated motor state variables, dual closed-loop vector control of speed and torque is performed on the motor, and a space vector pulse width modulation signal is generated to drive a three-phase full-bridge inverter connected to the permanent magnet synchronous motor, including: Obtain and calculate the target motor speed based on the motor speed command; Convert the motor's real-time angular velocity into real-time rotational speed; The difference between the target motor speed and the real-time speed is fed into the speed loop PI controller to obtain the reference values ​​of the d-axis and q-axis currents. The real-time values ​​of the d-axis and q-axis currents are compared with the reference values ​​of the d-axis and q-axis currents, respectively, and the difference is input to the torque loop PI controller to output the d-axis and q-axis reference voltages. The real-time values ​​of the d-axis and q-axis currents are obtained by converting the three-phase currents output to the permanent magnet synchronous motor by the three-phase inverter through Clark transformation and then undergoing coordinate transformation. After the d-axis and q-axis reference voltages are converted into α-axis and β-axis reference voltages through coordinate conversion, they are uniformly modulated by SVPWM to obtain the switching sequence control signals of the 6 IGBTs in the three-phase full-bridge inverter, so as to drive the three-phase full-bridge inverter connected to the permanent magnet synchronous motor.

8. The sensorless control method for permanent magnet synchronous motors based on improved extended Kalman filtering according to claim 7, characterized in that, The reference voltage component and the real-time acquired current component include the α-axis reference voltage component u. αref β-axis reference voltage component u βref α-axis current component i α β-axis current component i β ; Obtain the α-axis current component i α β-axis current component i β The methods include: Obtain the three-phase current output from the three-phase full-bridge inverter to the permanent magnet synchronous motor; Perform Clark transform on the three-phase currents to obtain the α-axis current component i. α β-axis current component i β ; Methods for obtaining the reference voltage component include: Obtain the d-axis and q-axis reference voltages output by the torque loop PI controller; Based on the d-axis and q-axis reference voltages, the coordinate transformation is used to obtain the α-axis and β-axis reference voltage components u. αref and β-axis reference voltage component u βref .

9. A sensorless control system for a permanent magnet synchronous motor based on an improved extended Kalman filter, characterized in that, include: Permanent magnet synchronous motor, three-phase full-bridge inverter, current sampling module, controller; The current sampling module is connected to the permanent magnet synchronous motor and is used to collect the three-phase current output from the three-phase full-bridge inverter to the permanent magnet synchronous motor and send it to the controller. The controller is connected to the three-phase full-bridge inverter and the current sampling module respectively, and is used to execute the control method as described in any one of claims 6-8, generate a space vector pulse width modulation signal, and output it to the three-phase full-bridge inverter, thereby driving the three-phase full-bridge inverter connected to the permanent magnet synchronous motor to operate.

Images