High-frequency injection synchronous full-parameter identification method and system based on permanent magnet synchronous motor position sensorless control

Abstract

This invention discloses a high-frequency injection synchronous full-parameter identification method and system based on sensorless control of a permanent magnet synchronous motor. The method includes: acquiring the motor's back electromotive force (EMF) signal using a sliding mode observer; extracting the fundamental frequency signal from the back EMF without phase delay using an adaptive synchronous frequency extractor; constructing a third-order phase-locked loop (PLL) by introducing angular acceleration as an additional state variable and then designing the feedback gain using pole placement; injecting high-frequency perturbation sinusoidal signals into the motor's d-q axes; reconstructing the observability matrix of the multi-parameter identification equation; and then achieving synchronous online identification of stator resistance, quadrature-axis and direct-axis inductance, and rotor flux linkage based on recursive least squares and extended Kalman filtering. This application eliminates the need for mechanical position sensors such as photoelectric encoders and rotary transformers, reducing system costs and simplifying the installation process. It supports applications in various scenarios, including new energy vehicle drives, industrial servos, and energy-saving home appliances. The algorithm can be implemented using a DSP / MCU.

Description

Technical Field

[0001] This invention belongs to the field of permanent magnet synchronous motor control technology, specifically relating to a high-frequency injection synchronization full parameter identification method and system based on sensorless control of permanent magnet synchronous motors. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs), with their significant advantages of small size, simple structure, low loss, high efficiency, and high power density, have been widely used in new energy vehicles, industrial servo systems, and energy-saving home appliances, becoming a core component of high-efficiency intelligent motor drives. In PMSM field-oriented control, rotor position information is a key parameter to ensure torque output efficiency and system stability, and the accuracy of motor parameters directly determines the performance of the control algorithm.

[0003] In traditional PMSM control, rotor position is usually obtained through mechanical position sensors such as photoelectric encoders and rotary transformers. Although these sensors can provide high-precision position feedback, they have inherent drawbacks: high-precision sensors are expensive, which significantly increases the system cost. It is susceptible to interference under harsh working conditions such as high temperature and vibration, resulting in a high failure rate and reduced system reliability. The installation process requires strict precision in areas such as coaxiality, which can easily introduce measurement errors and increase the complexity of the motor structure and the difficulty of installation.

[0004] To address the aforementioned issues, sensorless control technology has emerged, which estimates rotor position in real time through algorithms without the need for physical sensors. This has become a research hotspot in the field of motor control, and is particularly suitable for applications with urgent requirements for lightweight and high reliability.

[0005] Currently, sensorless position estimation techniques for PMSMs mainly fall into two major technical routes: The high-frequency signal injection method based on rotor salient pole characteristics: By injecting rotating high-frequency sinusoidal signals, pulsating high-frequency voltage signals, or high-frequency square wave signals into the motor, the high-frequency response generated by the rotor salient pole effect is used to extract rotor position information, which outperforms other methods under zero-speed or low-speed conditions. However, this method requires additional external signal injection, which inevitably introduces additional losses and errors to the motor, reducing operating efficiency. Furthermore, the rotating high-frequency sinusoidal signal injection method depends on the motor's salient pole and has a complex implementation process. The adaptability of the pulsating high-frequency voltage injection method to surface-mounted PMSMs still needs optimization. Although the high-frequency square wave injection method has a good dynamic response, it does not completely solve the position estimation error problem caused by inductance changes.

[0006] Model-based methods based on back EMF or flux linkage observation include algorithms such as Extended Kalman Filter (EKF), Sliding Mode Observer (SMO), and Model Reference Adaptive (MRAS). These methods estimate the back EMF or flux linkage by constructing a mathematical model of the motor, thereby deriving the rotor position and speed. Among them, the EKF algorithm considers model errors and measurement noise, resulting in high estimation accuracy, but it involves a large number of matrix operations, placing stringent requirements on processor computing power and real-time performance. The SMO algorithm is robust and has a fast response, but the discontinuity of the sign function can cause system chattering, leading to high-order harmonics in the estimated back EMF. Traditional low-pass filters (LPFs) introduce phase lag and amplitude attenuation when filtering out harmonics, further affecting the position estimation accuracy. The MRAS algorithm has a simple structure, but it is highly sensitive to motor parameters, and the design of adjustable models and adaptive laws is complex, limiting its applicability.

[0007] Among them, PMSM parameter identification technology is an important support for ensuring the accuracy of sensorless control. Motor parameters are easily affected by factors such as temperature changes and magnetic saturation effect, which can cause time-varying changes: temperature changes can lead to stator resistance deviation and permanent magnet flux floating, and even cause permanent demagnetization; magnetic saturation effect will cause the quadrature and direct axis inductance to change with the current, and the effect on q-axis inductance is more significant.

[0008] Existing parameter identification methods are divided into offline identification and online identification: Offline identification (volt-ampere method, finite element method, etc.) is easy to implement, but it does not consider the time-varying nature of parameters during operation, and temperature compensation methods require additional temperature sensors, increasing hardware costs; Online identification (recursive least squares method, extended Kalman filter method, etc.) can acquire parameters in real time, but it is constrained by the motor voltage equation and can only identify a maximum of two parameters at a time. It requires a step-by-step identification strategy, which is inefficient. Furthermore, under steady-state conditions, parameters such as direct-axis inductance and rotor flux linkage cannot be identified simultaneously, resulting in insufficient parameter coupling and identification accuracy, which in turn affects the robustness of sensorless control.

[0009] Furthermore, existing sensorless control technologies still have significant shortcomings under dynamic operating conditions: traditional second-order phase-locked loops (PLLs) ignore angular acceleration, resulting in sluggish dynamic response and increased tracking error when the rotational speed changes abruptly or the load increases suddenly; the accuracy of position estimation and the accuracy of parameter identification are coupled, and parameter perturbation can lead to the accumulation of position estimation errors, while the inaccuracy of position estimation will affect the reliability of parameter identification, making it difficult to meet the requirements of high-precision control under all operating conditions in scenarios such as new energy vehicle drive and high-speed variable load industrial servo.

[0010] In summary, existing sensorless PMSM control technology faces three core problems: time-varying parameters lead to insufficient accuracy in position estimation and parameter identification; lack of dynamic response capability makes it difficult to adapt to sudden changes in speed and load; and online identification of multiple parameters suffers from under-ranking problems, making it impossible to achieve synchronous and accurate identification of all parameters.

[0011] Therefore, a high-frequency injection synchronization full parameter identification method based on sensorless control of permanent magnet synchronous motor is proposed. Summary of the Invention

[0012] To address the problems existing in the prior art, this invention provides a high-frequency injection synchronous full parameter identification method and system based on sensorless control of permanent magnet synchronous motors. It eliminates the need for mechanical position sensors such as photoelectric encoders and rotary transformers, reducing system costs and simplifying the installation process. It supports multiple application scenarios such as new energy vehicle drives, industrial servo systems, and energy-saving home appliances. The algorithm can be implemented based on DSP / MCU, improving the overall performance of sensorless control of PMSMs.

[0013] To achieve the above objectives, the present invention provides the following solution: A high-frequency injection synchronization full parameter identification method based on sensorless control of permanent magnet synchronous motors, the method comprising: S1: The back EMF signal of the motor is obtained based on the sliding mode observer. The filter parameters are dynamically adjusted by the adaptive synchronous frequency extractor. Then, the fundamental component is extracted from the back EMF signal and an angle tracking closed loop is constructed by combining it with the orthogonal phase-locked loop. Finally, the influence of motor parameters on the phase-locking process is eliminated by normalization processing to realize the steady-state estimation of rotor position. S2: Based on the steady-state estimation of rotor position, a third-order phase-locked loop is constructed. By introducing angular acceleration as an additional state variable, and then using pole placement to design feedback gain, synchronous tracking of rotational speed and angular acceleration is achieved. S3: Based on the steady-state estimation of rotor position and the synchronous tracking of speed and angular acceleration, a high-frequency perturbation sinusoidal signal is injected into the dq axis of the motor. By reconstructing the observability matrix of the multi-parameter identification equation, the stator resistance, quadrature and direct axis inductance and rotor flux linkage are synchronously identified online based on the recursive least squares method and extended Kalman filter.

[0014] Preferably, the method for obtaining the back EMF signal of the motor based on the sliding mode observer includes: ; ; in, They are respectively shaft and Shaft stator voltage, They are respectively shaft and Shaft stator current, For stator resistance, For stator equivalent inductance, The back electromotive force component, To observe the current, This is a sliding mode switching control item.

[0015] Preferably, the method for dynamically adjusting filter parameters using an adaptive synchronization frequency extractor includes: ; in, The damping coefficient is... This is the estimated rotational speed.

[0016] Preferably, the method for achieving steady-state rotor position estimation by constructing an angle tracking closed loop using an orthogonal phase-locked loop and then eliminating the influence of motor parameters on the phase-locking process through normalization processing includes: The phase error is defined as: ; in, The fundamental back electromotive force; Based on the phase error, the estimated speed is obtained through the proportional-integral controller: ; Integrating the estimated rotational speed, we obtain the estimated rotor electrical angle: ; in, and These are the proportional gain and integral gain of the QPLL, respectively. The phase error is normalized, and the normalized phase error is expressed as: ; in, A small positive constant is set to prevent the denominator from being zero. After normalization, the input error of the phase-locked loop reflects the phase deviation rather than the amplitude fluctuation. In steady state, when At that time, there were: This enables steady-state zero-static-error estimation of the rotor position.

[0017] Preferably, the state variables of the third-order phase-locked loop include electrical angle, angular velocity, and angular acceleration, wherein the closed-loop poles are triple negative real roots. The closed-loop transfer function is , The feedback gain coefficient must satisfy the following conditions: , , ; Based on TOPLL transfer function This enables synchronous tracking of rotational speed and angular acceleration, correcting the output of the QPLL. This enables zero-static-error tracking of rotor position and speed under dynamic operating conditions.

[0018] Preferably, the high-frequency disturbance signal is a sinusoidal signal, which is injected into the motor. The axis makes the perturbation angular frequency satisfy ,in For the sampling period, make The shaft current contains periodic perturbation components, which makes the observability matrix of the identification equation full rank.

[0019] Preferably, the synchronization parameter identification based on the recursive least squares method is as follows: The recursive least squares (RLS) method with a forgetting factor is used to estimate the parameters online. Let the vector of parameters to be identified be: ; in, For resistance, For d-axis inductance, for Shaft inductor, For rotor flux linkage; Based on the vector of parameters to be identified, the regression equation is expressed as: ; in, For observation output, For the regression vector, To account for modeling errors, the recursive process of RLS is as follows: ; ; ; in, Forgetting factor, It is the covariance matrix; This is the gain matrix; This is the parameter estimation vector.

[0020] Preferably, the synchronization parameter identification based on extended Kalman filtering is as follows: ; in, The voltage along the d-axis. This is the q-axis voltage. For d-axis current, For q-axis current, It represents the electric angular velocity.

[0021] The present invention also provides a high-frequency injection synchronization full parameter identification system based on sensorless control of permanent magnet synchronous motor. The system is used to implement the aforementioned method and includes: a position estimation module, a dynamic tracking module, and a parameter identification module. The position estimation module is used to acquire the back EMF signal of the motor based on the sliding mode observer, dynamically adjust the filter parameters through the adaptive synchronous frequency extractor, extract the fundamental component from the back EMF signal, construct an angle tracking closed loop in combination with the orthogonal phase-locked loop, and finally eliminate the influence of motor parameters on the phase-locking process through normalization processing to realize the steady-state estimation of rotor position. The dynamic tracking module is used to construct a third-order phase-locked loop based on the steady-state estimation of the rotor position. By introducing angular acceleration as an additional state variable, and then using pole placement to design the feedback gain, synchronous tracking of rotational speed and angular acceleration is achieved. The parameter identification module is used to inject high-frequency perturbation sinusoidal signals into the dq axis of the motor based on the steady-state estimation of rotor position and the synchronous tracking of speed and angular acceleration. By reconstructing the observability matrix of the multi-parameter identification equation, the stator resistance, quadrature-direct axis inductance and rotor flux linkage are synchronously identified online based on the recursive least squares method and extended Kalman filtering.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows: According to the high-frequency injection synchronization full parameter identification method based on sensorless control of permanent magnet synchronous motor of the present invention, an adaptive synchronization frequency extractor is used to replace the traditional low-pass filter, and the filter parameters are dynamically adjusted to accurately extract the fundamental component, effectively suppressing high-order harmonic interference, avoiding phase lag and amplitude attenuation, and ensuring high fidelity of back electromotive force signal. By combining the normalization process of the orthogonal phase-locked loop, the influence of motor parameters on the phase-locking process is eliminated, and the steady-state accuracy is better than that of the traditional sliding mode observer. A third-order phase-locked loop is proposed, which introduces angular acceleration as an additional state variable, expands the system dimension, and optimizes the feedback gain through pole configuration to achieve synchronous tracking of rotational speed and angular acceleration. For dynamic operating conditions such as sudden changes in speed and sudden increases in load, it solves the problem of response lag caused by neglecting the acceleration term, significantly reduces the amplitude of speed fluctuations, and restores stability faster; By injecting high-frequency disturbance signals to reconstruct the observability matrix of the identification equation, the problem of the inability to simultaneously identify the direct-axis inductance and rotor flux under steady-state conditions is solved. The recursive least squares method and extended Kalman filtering are used to achieve synchronous online identification of all parameters of stator resistance, quadrature-axis and direct-axis inductance, and rotor flux, and to compensate for control deviations caused by time-varying parameters in real time. The observer optimizes the sliding mode gain based on Lyapunov stability analysis, balances the system convergence speed and chatter suppression, has strong anti-interference ability against parameter perturbation and electromagnetic interference, forms a closed loop of parameter identification and position estimation, corrects motor parameter drift in real time, ensures stable control performance under all operating conditions, and avoids risks such as excessive rotor position estimation error. It eliminates the need for mechanical position sensors such as photoelectric encoders and rotary transformers, reducing system costs and simplifying the installation process. It also enhances reliability under harsh working conditions, is compatible with surface-mount and built-in PMSMs, and supports applications in various scenarios such as new energy vehicle drives, industrial servos, and energy-saving home appliances. The algorithm can be implemented based on DSP / MCU. Attached Figure Description

[0023] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is an overall block diagram of a sliding mode observer sensorless control method based on a high-frequency injection synchronous full parameter identification method for sensorless control of a permanent magnet synchronous motor according to an embodiment 2 of the present invention. Figure 2 This is a block diagram of the TOPLL structure of a high-frequency injection synchronous full parameter identification method based on sensorless control of a permanent magnet synchronous motor according to an embodiment 2 of the present invention. Figure 3 This is a schematic flowchart of a high-frequency injection synchronization full parameter identification method based on sensorless control of a permanent magnet synchronous motor according to an embodiment 1 of the present invention. Detailed Implementation

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] The following describes, with reference to the accompanying drawings, a sensorless high-frequency injection parameter synchronization identification method for permanent magnet synchronous motors according to an embodiment of the present invention.

[0028] Example 1 like Figure 3As shown in the embodiment of the present invention, the high-frequency injection synchronous full-parameter identification method based on sensorless control of permanent magnet synchronous motor is presented. To further illustrate the difference between the present invention and the prior art, the inventiveness of the present invention lies not only in the separate use of sliding mode observer, adaptive synchronous frequency extractor, phase-locked loop and parameter identification algorithm, but also in the collaborative design of the above-mentioned technical means according to the link of "high-fidelity extraction of back EMF - synchronous tracking of rotor position and dynamic speed - high-frequency injection to reconstruct observability and realize online identification of full parameters". Specifically: First, an adaptive synchronous frequency extractor is used to replace the traditional low-pass filter to extract the fundamental wave of the back EMF signal output by the sliding mode observer, thereby reducing the phase lag and amplitude attenuation caused by the traditional low-pass filter; Second, angular acceleration is introduced as an additional state variable in the phase-locked loop to construct a third-order phase-locked loop to improve the dynamic tracking performance under sudden speed changes and sudden load changes; Third, by injecting high-frequency disturbance signals into the dq axis of the motor, the observability matrix of the multi-parameter identification equation is reconstructed, so that the stator resistance, quadrature-direct axis inductance and rotor flux linkage have the conditions for online synchronous identification. The above components are not isolated and superimposed, but rather form a closed-loop collaborative mechanism of position estimation, dynamic tracking, and parameter identification. This mechanism improves the accuracy and robustness of sensorless control under all operating conditions and may include the following steps: 1) In this embodiment, the three-phase stator voltage signal on the output side of the inverter is first acquired. and three-phase stator current signal The voltage and current signals in the three-phase stationary coordinate system are converted to the two-phase stationary coordinate system using Clark transformation. Below, get and .exist A voltage model of a permanent magnet synchronous motor is established in a coordinate system, and a sliding mode observer is constructed based on this model. By feedback correction of the error between the observed current and the actual current, the estimated value of the back electromotive force is obtained. Therefore, raw observation signals can be provided for subsequent fundamental wave extraction and rotor position estimation without the need for mechanical position sensors. The formula can be written as: in, They are respectively shaft and Shaft stator voltage, They are respectively shaft and Shaft stator current, For stator resistance, For stator equivalent inductance, This is the back electromotive force component. According to the above equation, it can be reconstructed using an observer. .

[0029] The sliding mode observer is constructed as follows: in, To observe the current, This is a sliding mode switching control term. The current observation error is defined as: The sliding mode switching option can then be: in, For sliding mode gain, For boundary layer thickness, This is a saturation function. The above equation allows the observed current to quickly approximate the actual current, and the back EMF component can be approximately reconstructed by the switching control term. Compared to traditional sliding mode observers that directly use the sign function, this embodiment suppresses high-frequency chattering by setting a boundary layer, providing a more stable back EMF input for subsequent fundamental wave extraction.

[0030] sliding mode gain Based on Lyapunov stability analysis, the following conditions must be met: ( ), ( )and ( )and correspond The constraints, among which The back electromotive force α Axial components; for Axis current observation error; For d-axis and q-axis inductance; Electric angular velocity; To achieve a balance between system convergence speed and chatter suppression in terms of stator resistance, the filter parameters are dynamically adjusted using an adaptive synchronization frequency extractor, and the sliding mode observer output... It contains higher harmonic components. To avoid the fixed phase lag introduced by traditional low-pass filters when extracting the fundamental frequency, this embodiment introduces an adaptive synchronous frequency extractor (ASFE), whose center frequency is adjusted in real time according to the estimated electrical angular frequency, thereby maintaining a high amplitude response at the fundamental frequency and achieving significant attenuation at higher harmonics. The closed-loop transfer function of the ASFE can be expressed as: in, The damping coefficient is... This is an estimated value for the rotational speed. Since the center frequency in the above equation is... The changes are synchronous, therefore, near the fundamental frequency, we have: And for For equal higher harmonic components, we have: Therefore, amplitude attenuation of third and higher harmonics can be achieved while maintaining the fundamental phase and amplitude essentially unchanged. The fundamental back electromotive force output by the ASFE is denoted as... This is for use by subsequent orthogonal phase-locked loops.

[0031] The fundamental back electromotive force output by ASFE Input a quadrature phase-locked loop (QPLL), construct the phase error, and establish an angle tracking closed loop. Specifically, the phase error is defined as: Based on this, the estimated speed value is obtained through the proportional-integral controller: By further integrating the estimated rotational speed, the estimated rotor electrical angle is obtained: in, and These are the proportional gain and integral gain of the QPLL, respectively. The above equation forms three parts: phase detection, loop filtering, and angle integration, thereby achieving closed-loop tracking of the rotor's electrical angle and electrical angular velocity.

[0032] To reduce the impact of back EMF amplitude variations and parameter perturbations on the phase-locked loop (PLL) process, this embodiment normalizes the phase error. The normalized phase error can be expressed as: in, A small positive constant is set to prevent the denominator from being zero. After normalization, the input error of the phase-locked loop mainly reflects the phase deviation rather than the amplitude fluctuation, thereby reducing the impact of motor parameter changes on position estimation. In steady state, when At that time, there were: Therefore, steady-state, error-free estimation of the rotor position can be achieved. If necessary, the initial angle estimate can also be obtained directly from the fundamental back electromotive force component. This is used as the initial value for the QPLL to improve locking speed.

[0033] 2) Construct a third-order phase-locked loop (PLL) by introducing angular acceleration as an additional state variable, and then design the feedback gain using pole placement. The state variables of the third-order PLL include electrical angle, angular velocity, and angular acceleration, and the closed-loop poles are triple negative real roots. ( The closed-loop transfer function is , Let be the feedback gain coefficient, and satisfy... , , This enables zero-static-error tracking of rotor position and speed, synchronous tracking of speed and angular acceleration, and improves dynamic response capability under sudden speed change conditions. 3) Inject a high-frequency disturbance signal into the motor, and reconstruct the observability matrix of the multi-parameter identification equation. The high-frequency disturbance signal is a sinusoidal or square wave signal, which is then injected into the motor. The axis makes the perturbation angular frequency satisfy ,in For the sampling period, make The shaft current contains periodic disturbance components, making the observability matrix of the identification equation full rank. Then, based on recursive least squares and extended Kalman filtering, the stator resistance, DC and quadrature axis inductances, and rotor flux linkage are synchronously identified online, thus solving the underrank problem of parameter identification under steady-state conditions. In the above process, the recursive least squares method adopts a recursive form with a forgetting factor, where the forgetting factor... The value ranges from 0.95 to 0.99. The recursive process includes gain matrix update, parameter estimation update and covariance matrix update. The synchronous parameter identification based on extended Kalman filter constructs a state vector containing stator current, inverse inductance and rotor flux linkage. It uses Jacobian matrix to linearize the nonlinear system and models the covariance matrix of process noise and measurement noise.

[0034] Meanwhile, the voltage and current signals in the three-phase stationary coordinate system are converted to the two-phase stationary coordinate system by the Clark transformation, and then to the two-phase rotating coordinate system by the Park transformation, thereby decoupling the mathematical model of the motor and providing a basis for position estimation and parameter identification.

[0035] Through the above steps, the speed estimation error, rotor position estimation error, and online identification parameter error under the condition of sudden speed change are significantly reduced compared with traditional algorithms.

[0036] Example 2 The effectiveness of this invention will be verified below with reference to specific implementation methods: To achieve high-precision, sensorless control of the engine speed from 0 to 3000 rpm across all operating conditions, the following method is adopted: First, the voltage and current signals in the three-phase stationary coordinate system are transformed to the two-phase stationary coordinate system using Clark transformation, specifically: in, Can represent voltage or current Then, the coordinates are transformed to a two-phase rotating coordinate system using the Park transformation to obtain the d-axis and q-axis components: The above formula can decouple the mathematical model of the motor, providing a basis for position estimation, speed tracking and parameter identification in sensorless control.

[0037] Initialize the controller parameters to enable the proportional gain of the current loop PI controller. Integral gain Speed ​​loop PI controller proportional gain Integral gain ; Initialize the parameters of the observation and identification module (such as...) Figure 1 Sliding Mode Observer (SMO) Sliding Mode Gain Adaptive Synchronous Frequency Extractor (ASFE) Parameters Initial value of fundamental frequency (Corresponding to an electrical angular velocity of 1000 rpm), third-order phase-locked loop (TOPLL) poles (Feedback gain) , , ); Collect the three-phase stator current on the output side of the inverter With voltage Converted to Clark transformation coordinate system and ; Substitute the observation equations of the sliding mode observer (SMO): Control input ,in, , The stator voltage components are in a two-phase stationary coordinate system. , The actual stator current components are shown in a two-phase stationary coordinate system. , The observed current component is the output of the sliding mode observer. , This is the switching control input for the sliding mode observer. For stator resistance, and They are respectively Shaft inductance and Shaft inductor, Electric angular velocity, For time variables, For sliding mode gain, For sign function, sliding mode gain The Lyapunov stability condition is satisfied: Ensure that the current observation error converges to 0; In this embodiment, the sampled Substituting into the discrete equations of the sliding mode observer, an update is performed once per control cycle. Its discrete form can be expressed as: in, The back electromotive force estimate can be reconstructed based on the switching control term: This yields the observed back electromotive force containing higher harmonics, which is then used as the input to the ASFE.

[0038] Will Given an ASFE structure, the ASFE closed-loop transfer function is: ( Dynamically extract the fundamental back electromotive force. Suppresses third and higher harmonics; The fundamental back electromotive force output by ASFE Input QPLL and calculate the phase error, speed estimate, and position estimate using the following steps: in, and These are proportional gain and integral gain, respectively. These are normalized small positive constants. Using the above formula, the estimated rotor position and rotational speed can be output. For example... Figure 2 As shown, the TOPLL state variables are expanded to (Electrical angle, angular velocity, angular acceleration), closed-loop pole configuration is triple negative real roots. Feedback gain according to formula , , Set as , , ; Based on TOPLL transfer function This enables synchronous tracking of rotational speed and angular acceleration, correcting the output of the QPLL. ; When the rotational speed changes abruptly, the rotational speed fluctuation amplitude of TOPLL is significantly reduced compared to that of traditional PLL, the recovery time is also shortened, and the dynamic response performance is significantly improved. High-frequency signal injection (e.g.) Figure 1 ),Towards -q axis injects a pulsed high-frequency voltage signal, with the following parameters: amplitude ,frequency (3.5% of the PWM switching frequency), satisfying approximate( , ); The Recursive Least Squares (RLS) algorithm with a forgetting factor is used. Initial covariance matrix Initial parameter estimates (Stator resistance) d-axis inductance Rotor flux ); In the synchronous parameter identification process, recursive least squares (RLS) with a forgetting factor is first used to estimate the parameters online. Let the vector of parameters to be identified be: The regression equation is expressed as: in, For observation output, For the regression vector, To account for modeling errors. The recursive process of RLS is as follows: in, The forgetting factor is preferably set to a value of [value to be filled in]. ; It is the covariance matrix; This is the gain matrix; This is the parameter estimation vector. Using the above formula, the parameter estimation result can be output once per control cycle; Constructing state vectors Observation vector ,in For d-axis current, For q-axis current, The reciprocal process noise covariance matrix of the d-axis inductance Measurement noise covariance matrix ; According to the EKF recursive formula in The voltage along the d-axis. For the q-axis voltage, the nonlinear system is linearized using the Jacobian matrix, and synchronous identification is performed. , , , .

[0039] The recursive least squares method has higher identification accuracy than the EKF method, with smaller parameter identification error, and solves the problem under steady-state conditions. and The underrank problem that cannot be identified simultaneously.

[0040] The rotor position estimate output by TOPLL Compared with the estimated speed Feedback to the Field Direction Control (FOC) module (such as...) Figure 1 In the case of surface-mounted permanent magnet synchronous motors, the preferred method is to use... If the control strategy is as follows, then the electromagnetic torque equation is: When taking At that time, the q-axis current setpoint can be determined by the target torque. The calculation yielded: Therefore, the FOC module generates d-axis and q-axis current setpoints based on the target torque or target speed, and combines them with the position estimate. The coordinate transformation and current closed-loop control are completed. If applied to an embedded permanent magnet synchronous motor, the maximum torque-to-current ratio (MTPA) strategy can be further determined. and .

[0041] Real-time parameters output by the parameter identification module ( The real-time parameters output by the parameter identification module This allows for online retuning of the current loop PI controller to compensate for parameter drift caused by temperature changes and magnetic saturation. Taking the d-axis current loop as an example, its PI controller parameters can be set using the bandwidth tuning method: Similarly, the PI parameters of the q-axis current loop can be expressed as: in, This is the cutoff angular frequency of the current loop. Therefore, when the stator resistance shifts due to temperature rise, or the quadrature-axis and direct-axis inductances change with operating conditions, the PI parameters can be updated in real time based on the identification results, thereby reducing control errors caused by parameter mismatch and improving system robustness. The inverter switching signal is generated by space vector pulse width modulation (SVPWM).

[0042] This embodiment 2 achieves high-precision and robust PMSM control without mechanical position sensors by using an adaptive synchronous frequency extractor (ASFE), sliding mode observer (SMO), third-order phase-locked loop (TOPLL), high-frequency injection and recursive least squares (RLS), and Kalman filter (EKF), thus adapting to the complex operating conditions required for new energy vehicle drives.

[0043] Example 3 The present invention also provides a high-frequency injection synchronous full parameter identification system based on sensorless control of permanent magnet synchronous motor. The system is used to implement the method described in Embodiment 1. The system includes: a position estimation module, a dynamic tracking module, and a parameter identification module. The position estimation module is used to acquire the back EMF signal of the motor based on the sliding mode observer, dynamically adjust the filter parameters through the adaptive synchronous frequency extractor, extract the fundamental component from the back EMF signal, and construct an angle tracking closed loop in combination with the orthogonal phase-locked loop. Finally, the influence of motor parameters on the phase-locking process is eliminated through normalization processing to realize the steady-state estimation of rotor position. The dynamic tracking module is used to construct a third-order phase-locked loop based on the steady-state estimation of rotor position. By introducing angular acceleration as an additional state variable, and then using pole placement to design feedback gain, synchronous tracking of rotational speed and angular acceleration is achieved. The parameter identification module is used to inject high-frequency perturbation sinusoidal signals into the dq axis of the motor based on the steady-state estimation of rotor position and the synchronous tracking of speed and angular acceleration. By reconstructing the observability matrix of the multi-parameter identification equation, the stator resistance, quadrature-direct axis inductance and rotor flux linkage are synchronously identified online based on the recursive least squares method and extended Kalman filtering.

[0044] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A high-frequency injection synchronization full-parameter identification method based on sensorless control of permanent magnet synchronous motors, characterized in that, The method includes: S1: The back EMF signal of the motor is obtained based on the sliding mode observer. The filter parameters are dynamically adjusted by the adaptive synchronous frequency extractor. Then, the fundamental component is extracted from the back EMF signal and an angle tracking closed loop is constructed by combining it with the orthogonal phase-locked loop. Finally, the influence of motor parameters on the phase-locking process is eliminated by normalization processing to realize the steady-state estimation of rotor position. S2: Based on the steady-state estimation of rotor position, a third-order phase-locked loop is constructed. By introducing angular acceleration as an additional state variable, and then using pole placement to design feedback gain, synchronous tracking of rotational speed and angular acceleration is achieved. S3: Based on the steady-state estimation of rotor position and the synchronous tracking of speed and angular acceleration, a high-frequency perturbation sinusoidal signal is injected into the dq axis of the motor. By reconstructing the observability matrix of the multi-parameter identification equation, the stator resistance, quadrature and direct axis inductance and rotor flux linkage are synchronously identified online based on the recursive least squares method and extended Kalman filter.

2. The method according to claim 1, characterized in that, Methods for obtaining the back EMF signal of a motor based on a sliding mode observer include: ； ； in, They are respectively shaft and Shaft stator voltage, They are respectively shaft and Shaft stator current, For stator resistance, For stator equivalent inductance, The back electromotive force component, To observe the current, This is a sliding mode switching control item.

3. The method according to claim 2, characterized in that, Methods for dynamically adjusting filter parameters using an adaptive synchronization frequency extractor include: ； in, The damping coefficient is... This is the estimated rotational speed.

4. The method according to claim 3, characterized in that, The method for achieving steady-state rotor position estimation by constructing an angle tracking closed loop using an orthogonal phase-locked loop and then eliminating the influence of motor parameters on the phase-locking process through normalization processing includes: The phase error is defined as: ； in, This is the fundamental back electromotive force; Based on the phase error, the estimated speed is obtained through the proportional-integral controller: ； Integrating the estimated rotational speed, we obtain the estimated rotor electrical angle: ； in, and These are the proportional gain and integral gain of the QPLL, respectively. The phase error is normalized, and the normalized phase error is expressed as: ； in, A small positive constant is set to prevent the denominator from being zero. After normalization, the input error of the phase-locked loop reflects the phase deviation rather than the amplitude fluctuation. In steady state, when At that time, there were: This enables steady-state zero-static-error estimation of the rotor position.

5. The method according to claim 4, characterized in that, The state variables of a third-order phase-locked loop include electrical angle, angular velocity, and angular acceleration, with the closed-loop poles being triple negative real roots. The closed-loop transfer function is , The feedback gain coefficient must satisfy the following conditions: , , ; Based on TOPLL transfer function This enables synchronous tracking of rotational speed and angular acceleration, correcting the output of the QPLL. This enables zero-static-error tracking of rotor position and speed under dynamic operating conditions.

6. The method according to claim 5, characterized in that, The high-frequency disturbance signal is a sinusoidal signal, which is injected into the motor. The axis makes the perturbation angular frequency satisfy ,in For the sampling period, make The shaft current contains periodic perturbation components, which makes the observability matrix of the identification equation full rank.

7. The method according to claim 6, characterized in that, The synchronization parameters identified based on the recursive least squares method are as follows: The recursive least squares (RLS) method with a forgetting factor is used to estimate the parameters online. Let the vector of parameters to be identified be: ； in, For resistance, For d-axis inductance, for Shaft inductor, For rotor flux linkage; Based on the vector of parameters to be identified, the regression equation is expressed as: ； in, For observation output, For the regression vector, To account for modeling errors, the recursive process of RLS is as follows: ； ； ； in, Forgetting factor, It is the covariance matrix; This is the gain matrix; This is the parameter estimation vector.

8. The method according to claim 7, characterized in that, The synchronization parameters based on the extended Kalman filter are identified as follows: ； in, The voltage along the d-axis. This is the q-axis voltage. For d-axis current, For q-axis current, ω is the electric angular velocity.

9. A high-frequency injection synchronization full-parameter identification system based on sensorless control of a permanent magnet synchronous motor, the system being used to implement the method described in any one of claims 1-8, characterized in that, The system includes: a position estimation module, a dynamic tracking module, and a parameter identification module; The position estimation module is used to acquire the back EMF signal of the motor based on the sliding mode observer, dynamically adjust the filter parameters through the adaptive synchronous frequency extractor, extract the fundamental component from the back EMF signal, construct an angle tracking closed loop in combination with the orthogonal phase-locked loop, and finally eliminate the influence of motor parameters on the phase-locking process through normalization processing to realize the steady-state estimation of rotor position. The dynamic tracking module is used to construct a third-order phase-locked loop based on the steady-state estimation of the rotor position. By introducing angular acceleration as an additional state variable, and then using pole placement to design the feedback gain, synchronous tracking of rotational speed and angular acceleration is achieved. The parameter identification module is used to inject high-frequency perturbation sinusoidal signals into the dq axis of the motor based on the steady-state estimation of rotor position and the synchronous tracking of speed and angular acceleration. By reconstructing the observability matrix of the multi-parameter identification equation, the stator resistance, quadrature-direct axis inductance and rotor flux linkage are synchronously identified online based on the recursive least squares method and extended Kalman filtering.

Images