Full-speed-domain position-sensorless control method for permanent magnet synchronous motor

Abstract

The invention discloses a full-speed-domain position-sensorless control method for a permanent magnet synchronous motor, relates to the technical field of motor driving control, and designs a nonlinear third-order extended state observer based on a hyperbolic tangent function to replace a PI regulator in a traditional phase-locked loop by adopting a pulsating high-frequency voltage injection method in a low-speed domain. System disturbance is estimated and compensated in real time through a third-order observer, and the angle observation precision and robustness under the dynamic working condition are improved; in a high-speed domain, a sliding-mode observer is adopted, a second-order generalized integrator is introduced to replace a traditional low-pass filter to filter a sliding-mode control law, a back electromotive force signal without phase lag is output, a phase-locked loop is cascaded to demodulate a rotor position and a rotating speed, and high-frequency jitter and phase delay are eliminated; in the full-speed-domain transition stage, a weighted switching strategy is designed, a switching interval is judged through a state machine, angle fusion is conducted through a trigonometric function vector synthesis method, the problem of periodic jumping of the angle is solved, and full-speed-domain smooth operation is achieved.

Description

A sensorless control method for permanent magnet synchronous motors across the entire speed range Technical Field

[0001] This invention relates to the field of motor drive control technology, and in particular to a sensorless control method for a permanent magnet synchronous motor across the entire speed range. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) are widely used in aerospace, electric vehicles, and industrial automation due to their high power density, high efficiency, and excellent dynamic performance. In many applications, sensorless control technology has become a research hotspot to improve system reliability, reduce costs, and minimize size.

[0003] Current sensorless control technologies are mainly divided into two categories: one is the high-frequency injection method (HFI) suitable for the zero-speed domain; the other is the observer method based on the back-EMF model, such as the sliding mode observer (SMO), suitable for the medium-speed domain. However, existing technologies have the following problems in practical applications: 1. Insufficient performance of traditional PLLs in the low-speed domain: Traditional HFI methods obtain a position error signal containing rotor position information through signal demodulation, and usually use a phase-locked loop (PLL) based on a PI controller for tracking. However, the PI controller is essentially a linear controller with a fixed bandwidth. At startup, the system is easily affected by speed fluctuations and high-frequency injection ripple. Traditional PI-PLLs struggle to balance dynamic response speed and noise suppression capabilities, resulting in large fluctuations in the estimated angle at low speeds, or even loss of lock.

[0004] 2. Phase lag and jitter in high-speed domain SMO: Traditional sliding mode observers use a switching function (Sign) to force the observed current to track the actual current, resulting in a back EMF control quantity containing severe high-frequency jitter. Existing technologies typically cascade a low-pass filter (LPF) to filter out high-frequency noise, but this introduces phase lag that varies with rotational speed, requiring complex angle compensation. Furthermore, the compensation accuracy is greatly affected by parameters, limiting control performance at high speeds.

[0005] 3. Poor smoothness of full-speed domain switching: Simple threshold switching can easily cause angle jumps in the transition range, resulting in current surges and affecting the stability of system operation.

[0006] To address the aforementioned issues, there is an urgent need for a sensorless control method for permanent magnet synchronous motors across the entire speed range. Summary of the Invention

[0007] To address the aforementioned issues, this application proposes a sensorless control method for permanent magnet synchronous motors across the entire speed domain based on third-order NLESO-PLL and SOGI-SMO. This method aims to solve the problems of poor low-speed disturbance rejection, high-speed phase lag, and speed fluctuations in traditional methods, while maintaining high-precision rotor position observation even under dynamic speed changes.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A sensorless control method for a permanent magnet synchronous motor in the full speed domain, comprising the following steps: S1, estimating the axis of the synchronous rotating coordinate system... Inject a high-frequency square wave voltage signal to extract the high-frequency current envelope containing position error. S2. A third-order nonlinear extended state observer is constructed using a smooth and continuous hyperbolic tangent function as the nonlinear error feedback law. This third-order nonlinear extended state observer is then used to monitor the high-frequency current envelope. The angle of the rotor in the low-speed range is obtained through processing. With speed S3. An improved sliding mode observer (SMO) model is built by replacing the low-pass filter (LPF) with a second-order generalized integrator (SOGI). The improved SMO model is then used to process the voltage and current of the α-β axes of the permanent magnet synchronous motor to obtain the pure back electromotive force. S4, the pure back electromotive force The rotor angle is obtained by normalizing and demodulating the input through a back-EMF phase-locked loop. With speed S5. A soft switching strategy based on hysteresis logic and vector synthesis is established to smoothly fuse observation data from the low-speed and high-speed domains. The soft switching strategy uses vector synthesis to convert low-speed and high-speed angles into vector components on the unit circle, which are then weighted and synthesized before being restored to angles, thus mathematically avoiding the problem of periodic angle jumps. The soft switching strategy receives rotor angles from the low-speed domain. With speed The angle of the high-speed domain rotor With speed And the current feedback speed As input, after state judgment and weighted calculation, the final control angle is output. With speed .

[0009] Preferably, in S1, the axes of the synchronous rotating coordinate system are estimated. Inject a high-frequency square wave voltage signal to extract the high-frequency current envelope containing position error. The specific content includes: estimating the synchronous rotating coordinate system The expression for injecting a high-frequency square wave voltage signal into the shaft is: ;in, ;in, To inject amplitude, For the injection signal period, The number of injection signal cycles, for High-frequency voltage components of the shaft. for High-frequency voltage components of the shaft. To inject a high-frequency voltage signal, The high-frequency voltage injection time point is used; the FOC algorithm is used to drive the motor rotation, and the three-phase current of the motor is obtained through current acquisition and Clarke transformation. The current values ​​of the two axes are used to extract the high-frequency current envelope containing the position error using a demodulation algorithm. The expression for the demodulation algorithm is: ;in, This indicates the estimated rotor electrical angle. They represent the extracted High-frequency current components of the shaft, This represents the inductance values ​​along the d-axis and q-axis. Indicates the electrical angle error value. This represents the switching factor corresponding to the polarity of the high-frequency square wave injection. An integer index that increments over time. The sampling period.

[0010] Preferably, the specific content of constructing the third-order nonlinear extended state observer in S2 includes: modeling the error tracking system PLL as a disturbed second-order system, and defining the state variables of the second-order system: ;in, For angular error, For speed error, This is an extended state, containing unmodeled dynamics and external disturbances. Indicates the angle error value. This represents the total external disturbance. Represent the derivative of the angle error; construct a discretized third-order nonlinear extended state observer, expressed as: ;in, System status Observed values; Input the error signal demodulated by HFI; The sampling period; The observer gain determines the observation bandwidth; It is the hyperbolic tangent function. To adjust the steepness of the nonlinear function, The observation error at each sampling point, Indicates the sampling point number.

[0011] The calculated state The predicted rotational speed is obtained by performing one integration, and the predicted angle is obtained by performing another integration.

[0012] Preferably, the specific content of building the improved sliding mode observer (SMO) model in S3 includes: defining a two-phase stationary coordinate system. The current state equation of the PMSM is as follows: Constructing a sliding mode observer: The sliding mode control signal is as follows: ;in, Indicates sliding mode gain. Represents a symbolic function. These represent the actual current value and the estimated current value, respectively. For motor phase resistance, For motor inductance, For motor The magnitude of the voltage on the shaft, This is the actual back electromotive force value of the motor. Predict the back electromotive force value for the motor; Includes back EMF information and high-frequency switching noise; processed using a second-order generalized integrator SOGI. The transfer function for SOGI is: ;in, Let be the transfer function of the second-order generalized integrator. This is the damping coefficient. It is used to adjust the filtering bandwidth of SOGI and the dynamic response speed of the system. For the Laplace operator; the center frequency of SOGI The estimated electric angular velocity is set in real time as the final overall solution output. Output after SOGI processing This is the fundamental back electromotive force.

[0013] Preferably, the position error in the back EMF phase-locked loop is defined as: ;in, These represent the fundamental back electromotive force values, This represents the electrical angle value predicted by the SMO. The position error signal is predicted by the SMO; this error is passed through a PI controller to output the rotational speed, and the angle is obtained by integration. .

[0014] Preferably, the specific content of building the soft switching strategy based on hysteresis logic and vector synthesis in S5 includes: S501, constructing a hysteresis state mechanism based on speed feedback to improve the stability of the transition zone, and outputting the state interval to which the current speed belongs based on the hysteresis state mechanism; S502, calculating the fusion weight coefficient G based on the output result of the hysteresis state mechanism and the current speed to form a smooth transition slope between the low-speed zone and the high-speed zone; S503, using the vector synthesis method for angle fusion, and using the four-quadrant arctangent function to synthesize the vector. The final fusion angle is obtained by performing calculations. The final output speed is obtained by calculating using a conventional linear weighted formula. .

[0015] Preferably, the specific content of the hysteresis state mechanism based on speed feedback in S501 includes: preset three key speed thresholds, and low-speed cutoff speed. Switch to high-speed mode speed and the speed when switching to high-speed mode And satisfy Define a persistent state variable to indicate whether the current state is in pure high-speed mode. The specific logic judgment process is as follows: If the current state is "pure high-speed mode", the decision is made only when the absolute value of the feedback speed is reached. Fall to The state will only be switched to false and exit pure high-speed mode when the following conditions are met; otherwise, if the current state is "mixed / low-speed mode", the state will only be switched to false and exit pure high-speed mode when the absolute value of the feedback speed is reached. Exceed Only when the state is set to true will it enter pure high-speed mode.

[0016] Preferably, the specific content of S502 includes: wherein, the fusion weighting coefficient The value ranges from 0 to 1, where 0 represents using HFI data exclusively and 1 represents using SMO data exclusively. When the state machine indicates that it is in "pure high-speed mode", the weighting coefficient is forcibly adjusted. Set to 1.0 to directly transmit high-speed observation data to save computing resources; when the state machine indicates that it is in non-high-speed mode, perform linear interpolation calculation based on the real-time rotational speed: if the absolute value of the current rotational speed is less than or equal to... ,but Set to 0.0; if the absolute value of the current speed is within... and Between, then weight According to the formula Perform the calculation.

[0017] Preferably, the specific content of angle fusion using the S503 vector synthesis method includes: using weights to combine low-speed angles... With high speed angle Map each component to a weighted vector component on the unit circle to construct the x and y coordinates of the composite vector: Where G is the fusion weight coefficient; the synthesized vector is synthesized using the four-quadrant arctangent function. The final fusion angle is obtained by performing calculations. .

[0018] The preferred expression for the conventional linear weighted formula is: ;in, This represents the rotational speed value calculated by the high-frequency injection method in the low-speed domain. This represents the rotational speed value calculated by the high-speed domain sliding mode observer method. Indicates the fusion weight coefficient. This represents the output rotational speed value after weighting.

[0019] In summary, the sensorless control method for a permanent magnet synchronous motor across the entire speed domain of this invention has the following advantages compared to traditional technologies: 1. The proposed third-order NLESO uses a hyperbolic tangent function, which has stronger disturbance observation and compensation capabilities compared to traditional PI or linear observers; 2. By using SOGI instead of the traditional LPF, the phase lag problem caused by filtering is alleviated in principle, eliminating the cumbersome phase compensation calibration process, improving the angle observation accuracy at high speeds, and reducing speed fluctuations; 3. The combination of the filtering characteristics of NLESO and the bandpass characteristics of SOGI makes the speed and position signal waveforms smooth across the entire speed domain, significantly reducing sliding mode jitter; 4. The angle prediction of the improved method can closely match the actual angle with only a small angle lag.

[0020] The technical method of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] Figure 1 is a schematic diagram of the overall solution process; Figure 2 is a schematic diagram of the overall simulation setup; Figure 3 is a schematic diagram of the high-frequency injection module; Figure 4 is a schematic diagram of the sliding mode observer module; Figure 5 is a schematic diagram of the high-speed and low-speed domain switching module; Figure 6 is a schematic diagram of the low-speed section speed following comparison, where (a) in Figure 6 is the overall schematic diagram and (b) in Figure 6 is an enlarged schematic diagram of point A in (a); Figure 7 is a schematic diagram of the low-speed section speed error comparison; Figure 8 is a schematic diagram of the high-speed section speed following comparison, where (a) in Figure 8 is the overall schematic diagram and (b) in Figure 8 is an enlarged schematic diagram of point B in (a); Figure 9 is a schematic diagram of the high-speed section speed error comparison; Figure 10 is a schematic diagram of the overall speed following comparison; Figure 11 is a schematic diagram of the overall speed error comparison; Figure 12 is a schematic diagram of the improved method angle following, where (a) in Figure 12 is the overall schematic diagram and (b) in Figure 12 is an enlarged schematic diagram of point C in (a). Detailed Implementation

[0022] The technical method of the present invention will be further described below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application.

[0023] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the scope of this application and its application or use.

[0024] Techniques, systems, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, they should be considered part of the instruction manual.

[0025] In all the examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0026] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0027] A sensorless control method for a permanent magnet synchronous motor across the entire speed domain, as shown in Figure 1, includes the following steps: S1, estimating the axes of the synchronous rotating coordinate system... Inject a high-frequency square wave voltage signal to extract the high-frequency current envelope containing position error. .

[0028] S2. A third-order nonlinear extended state observer is constructed using a smooth and continuous hyperbolic tangent function as the nonlinear error feedback law. This third-order nonlinear extended state observer is then used to measure the high-frequency current envelope. The angle of the rotor in the low-speed range is obtained through processing. With speed .

[0029] S3. An improved sliding mode observer (SMO) model is built by replacing the low-pass filter (LPF) with a second-order generalized integrator (SOGI). The pure back electromotive force is obtained by processing the voltage and current of the α-β axes of the permanent magnet synchronous motor using the improved SMO model. .

[0030] S4, The pure back electromotive force The rotor angle is obtained by normalizing and demodulating the input through a back-EMF phase-locked loop. With speed .

[0031] S5. Develop a soft switching strategy based on hysteresis logic and vector synthesis to smoothly fuse observation data from the low-speed and high-speed domains.

[0032] The soft switching strategy employs a vector synthesis method, which converts low-speed and high-speed angles into vector components on a unit circle, weights them together, and then restores them to angles, thus mathematically avoiding the problem of periodic angle jumps.

[0033] The soft switching strategy receives the low-speed domain rotor angle. With speed The angle of the high-speed domain rotor With speed And the current feedback speed As input, after state judgment and weighted calculation, the final control angle is output. With speed .

[0034] Example 1 Low-speed domain: Third-order nonlinear expanding state observer (NLESO-PLL) based on hyperbolic tangent function.

[0035] To address the poor disturbance rejection capability of traditional PI-PLL systems, this invention introduces the concept of active disturbance rejection control. All factors, including speed fluctuations, injected ripple, and model mismatch, are defined as the "total disturbance," and a third-order nonlinear extended state observer is constructed.

[0036] Unlike traditional Expanding State Observers (ESOs) and improved fal functions, this invention employs a smooth and continuous hyperbolic tangent function. As a nonlinear error feedback law The function exhibits linearity near the origin and saturation when the error is large. This avoids oscillations caused by high gain and ensures fast convergence. NLESO directly replaces the PI element in the traditional PLL, achieving zero steady-state error tracking of position errors and real-time compensation for disturbances.

[0037] Furthermore, in S1, the axes of the synchronously rotating coordinate system are estimated. Inject a high-frequency square wave voltage signal to extract the high-frequency current envelope containing position error. The specific content includes: estimating the synchronous rotating coordinate system The expression for injecting a high-frequency square wave voltage signal into the shaft is: .

[0038] in, .

[0039] in, To inject amplitude, For the injection signal period, The number of injection signal cycles, for High-frequency voltage components of the shaft. for High-frequency voltage components of the shaft. To inject a high-frequency voltage signal, This refers to the high-frequency voltage injection time point.

[0040] Because the square wave injection frequency is high, there is no need for a complex bandpass filter. The FOC algorithm is used to drive the motor rotation. The three-phase current of the motor is acquired through current acquisition and then transformed by Clarke. The current values ​​of the two axes are used to extract the high-frequency current envelope containing the position error using current difference or demodulation algorithms. The expression for the demodulation algorithm is: .

[0041] in, This indicates the estimated rotor electrical angle. They represent the extracted High-frequency current components of the shaft, This represents the inductance values ​​along the d-axis and q-axis. Indicates the electrical angle error value. This represents the switching factor corresponding to the polarity of the high-frequency square wave injection. An integer index that increments over time. The sampling period.

[0042] Traditional PLLs typically use a PI controller to regulate the circuit, bringing it towards zero. Their closed-loop transfer function is equivalent to a second-order system, and under dynamic conditions, the integral element is prone to saturation or response lag.

[0043] Furthermore, the specific content of constructing the third-order nonlinear extended state observer in S2 includes: modeling the error tracking system PLL as a disturbed second-order system, and defining the state variables of the second-order system: .

[0044] in, For angular error, For speed error, This is an extended state, containing unmodeled dynamics and external disturbances. Indicates the angle error value. This represents the total external disturbance. This represents the differential of the angular error.

[0045] The discretized third-order nonlinear extended state observer (NLESO) is constructed, i.e., the discretized third-order NLESO observer, and its expression is: .

[0046] in, System status Observed values; Input the error signal demodulated by HFI; The sampling period; The observer gain determines the observation bandwidth; It is the hyperbolic tangent function. To adjust the steepness of the nonlinear function, The observation error at each sampling point, Indicates the sampling point number.

[0047] The calculated state The predicted rotational speed is obtained by performing one integration, and the predicted angle is obtained by performing another integration.

[0048] use The function in error When the error is large, the output tends to saturate, preventing system overshoot caused by traditional high gain; when the error is small, it has high gain characteristics, ensuring fast convergence.

[0049] By using the observed disturbance state for feedforward compensation, the PLL can still effectively suppress the inherent second harmonic ripple of the HFI under low damping conditions such as no load, and output a stable estimated speed. Finally, the estimated speed and angle are derived from the observer state and obtained by integration.

[0050] Example 2 High-speed domain: An improved sliding mode observer (SMO) based on the second-order generalized integrator-phase-locked loop (SOGI-PLL).

[0051] To address the phase lag problem caused by traditional SMO plus a traditional low-pass filter (LPF), this invention uses a second-order generalized integrator to replace the LPF.

[0052] SOGI has frequency selectivity; when its center frequency is set to the synchronous electrical angular velocity, it can extract the fundamental component of the back electromotive force in the sliding mode control law without attenuation or phase lag, while effectively filtering out high-order harmonics introduced by the sliding mode switch. The purified back electromotive force after SOGI filtering is fed into the phase-locked loop for normalization and angle demodulation, thereby obtaining a high-precision rotor position.

[0053] Furthermore, the specific details of building the improved sliding mode observer (SMO) model in S3 include: defining a two-phase stationary coordinate system. The current state equation of the PMSM is as follows: .

[0054] Constructing a sliding mode observer: .

[0055] Among them, the sliding mode control signal is: .

[0056] in, Indicates sliding mode gain. Represents a symbolic function. These represent the actual current value and the estimated current value, respectively. For motor phase resistance, For motor inductance, For motor The magnitude of the voltage on the shaft, This is the actual back electromotive force value of the motor. Predict the back electromotive force value for the motor.

[0057] It includes back EMF information and high-frequency switching noise.

[0058] Traditional methods use low-pass filters to extract back electromotive force, which introduces phase lag. This invention uses a second-order generalized integrator (SOGI) for processing. The transfer function for SOGI is: .

[0059] in, Let be the transfer function of the second-order generalized integrator. This is the damping coefficient. It is used to adjust the filtering bandwidth of SOGI and the dynamic response speed of the system. For the Laplace operator.

[0060] The center frequency of SOGI The estimated electric angular velocity is set in real time as the final overall solution output. .

[0061] Because SOGI at the resonant frequency The amplitude gain at point A is 1 and the phase lag is 0. Therefore, the output after SOGI processing is... It is the fundamental back electromotive force, and there is no phase lag, which eliminates the phase compensation link that varies with the rotational speed and greatly simplifies the algorithm.

[0062] Obtain a pure back electromotive force Instead of using the arctangent function to calculate the angle (to avoid differential noise), a back EMF phase-locked loop is constructed.

[0063] Furthermore, the position error in the back EMF phase-locked loop is defined as: .

[0064] in, These represent the fundamental back electromotive force values, respectively. This represents the electrical angle value predicted by the SMO. This is the position error signal predicted by the SMO.

[0065] This error is passed through a PI controller, which outputs the rotational speed; the angle is obtained by integration. .

[0066] Example 3 Switching Strategy: Hysteresis Logic + Vector Synthesis.

[0067] A state machine-based hysteresis switching logic was designed to prevent repeated switching near the threshold. In the angle fusion algorithm, direct algebraic weighting was abandoned in favor of vector synthesis: low-speed and high-speed angles were converted into vector components on the unit circle, weighted, synthesized, and then restored to angles, thus mathematically avoiding the problem of periodic angle jumps.

[0068] This invention designs a soft switching strategy based on hysteresis logic and vector synthesis to achieve smooth fusion of low-speed observation data and high-speed observation data.

[0069] Furthermore, the specific content of building a soft switching strategy based on hysteresis logic and vector synthesis in S5 includes: S501, In order to prevent the control mode from frequently changing due to signal noise when the motor is running near the switching threshold, this embodiment constructs a hysteresis state mechanism based on speed feedback to improve the stability of the transition zone, and outputs the state interval to which the current speed belongs based on the hysteresis state mechanism.

[0070] Furthermore, the specific details of the hysteresis state mechanism based on speed feedback in S501 include: presetting three key speed thresholds, and a low-speed cutoff speed. Switch to high-speed mode speed and the speed when switching to high-speed mode And satisfy .

[0071] Define a persistent state variable to indicate whether the current state is in pure high-speed mode. The specific logic judgment process is as follows: If the current state is "pure high-speed mode", the system will check if the absolute value of the feedback speed is reached. Fall to Only when the following conditions are met will the status be switched to false, exiting the pure high-speed mode.

[0072] Conversely, if the current mode is "mixed / low speed", only when the absolute value of the feedback speed is... Exceed Only when the state is set to true will it enter pure high-speed mode.

[0073] By establishing a hysteresis interval, the stability of the system in the transition region is effectively improved.

[0074] S502. Based on the output results of the hysteresis state mechanism and the current rotation speed, calculate the fusion weight coefficient G to form a smooth transition slope between the low-speed region and the high-speed region.

[0075] Furthermore, the specific content of S502 includes: among which, the fusion weight coefficient The value ranges from 0 to 1, where 0 represents using HFI data exclusively and 1 represents using SMO data exclusively.

[0076] When the state machine indicates that it is in "pure high-speed mode", the weighting coefficients are forcibly adjusted. Set to 1.0 to directly transmit high-speed observation data to save computing resources.

[0077] When the state machine indicates that it is in non-high-speed mode, linear interpolation calculation is performed based on the real-time speed: if the absolute value of the current speed is less than or equal to... ,but Take 0.0.

[0078] If the absolute value of the current speed is located at and Between, then weight According to the formula Perform the calculation.

[0079] S503. To address the periodic jump problem in the angle fusion process, this invention abandons the traditional algebraic weighting method and adopts a vector synthesis method for angle fusion, utilizing the four-quadrant arctangent function to synthesize the vector. The final fusion angle is obtained by performing calculations. .

[0080] The final output speed is obtained by calculating using a conventional linear weighted formula. .

[0081] Furthermore, the specific details of angle fusion using the S503 vector synthesis method include: utilizing weights to fuse low-speed angles. With high speed angle Map each component to a weighted vector component on the unit circle to construct the x and y coordinates of the composite vector: .

[0082] Where G is the fusion weight coefficient.

[0083] Using the four-quadrant arctangent function to synthesize vectors The final fusion angle is obtained by performing calculations. This method, from a mathematical perspective, guarantees the continuity of the synthesized angle throughout the entire circumference.

[0084] Furthermore, the expression for the conventional linear weighted formula is: .

[0085] in, This represents the rotational speed value calculated by the high-frequency injection method in the low-speed domain. This represents the rotational speed value calculated by the high-speed domain sliding mode observer method. Indicates the fusion weight coefficient. This represents the output rotational speed value after weighting.

[0086] Example 4 verifies the method by building a speed-current dual closed-loop sensorless control system for a permanent magnet synchronous motor using MATLAB / Simulink. High-frequency injection combined with a third-order NLESO-PLL is used for the low-speed range, while the SGIO-SMO method is used for the high-speed range. Since the permanent magnet synchronous motor undergoes a calibration process 0.2 seconds before system startup, the expected speed is set to increase from 0.2 seconds to 200 rpm, maintain this speed for 0.8 seconds, then continue increasing to 1000 rpm and maintaining this speed. The overall simulation time is set to 2.5 seconds. The overall simulation setup is shown in Figures 2-5. In Figure 5... This indicates the sampling delay, meaning that the data sampled at the previous time step is used.

[0087] The simulation results are shown in Figures 6 and 7. When the speed starts to increase, the method of the present invention can effectively reduce the speed fluctuation and quickly converge to near the actual speed. The speed error during the dynamic speed change segment can be kept within 10 rpm, and the speed during the stable speed segment can be kept within 5 rpm. It can be seen that the third-order NLESO proposed in this application adopts the hyperbolic tangent function, which has a stronger disturbance observation and compensation capability compared with the traditional PI or ESO.

[0088] As shown in Figures 8 and 9, this method can significantly reduce the predicted speed fluctuation of SMO and keep the speed error in the high-speed section within 10 rpm. It can be seen that by using SOGI to replace the traditional LPF, the phase lag problem caused by filtering is alleviated in principle, the cumbersome phase compensation calibration process is eliminated, the angle observation accuracy in the high-speed section is improved, and the speed fluctuation is reduced.

[0089] The simulation results are shown in Figures 10 and 11. The combination of NLESO's filtering characteristics and SOGI's bandpass characteristics makes the speed and position signal waveforms smooth across the entire speed domain, significantly reducing sliding mode jitter. The hysteresis logic-based state machine and linear interpolation calculation fusion weights ensure the smooth fusion of the two algorithms in the transition zone, avoiding current oscillations caused by control mode switching. The speed dynamic change segment error is kept within 10 rpm, and the speed steady segment error can be kept within 5 rpm.

[0090] As shown in Figure 12, the angle prediction using the improved method of the present invention can closely match the reference angle with only a small angle lag.

[0091] Finally, it should be noted that the above embodiments are only used to illustrate the technical methods of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical methods of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical methods to deviate from the spirit and scope of the technical methods of the present invention.

Claims

1. A sensorless control method for a permanent magnet synchronous motor across its entire speed range, characterized in that, Includes the following steps: S1, estimating the axes of the synchronous rotating coordinate system Inject a high-frequency square wave voltage signal to extract the high-frequency current envelope containing position error. ； S2. A third-order nonlinear extended state observer is constructed using a smooth and continuous hyperbolic tangent function as the nonlinear error feedback law. This third-order nonlinear extended state observer is then used to measure the high-frequency current envelope. The angle of the rotor in the low-speed range is obtained through processing. With speed S3. An improved sliding mode observer (SMO) model is built by replacing the low-pass filter (LPF) with a second-order generalized integrator (SOGI). The improved SMO model is then used to process the voltage and current of the α-β axes of the permanent magnet synchronous motor to obtain the pure back electromotive force. ； S4, The pure back electromotive force The rotor angle is obtained by normalizing and demodulating the input through a back-EMF phase-locked loop. With speed S5. A soft switching strategy based on hysteresis logic and vector synthesis is established to smoothly fuse observation data from the low-speed and high-speed domains. The soft switching strategy uses vector synthesis to convert low-speed and high-speed angles into vector components on the unit circle, which are then weighted and synthesized before being restored to angles, thus mathematically avoiding the problem of periodic angle jumps. The soft switching strategy receives rotor angles from the low-speed domain. With speed The angle of the high-speed domain rotor With speed And the current feedback speed As input, after state judgment and weighted calculation, the final control angle is output. With speed 。 2. The sensorless control method for a permanent magnet synchronous motor across the entire speed range according to claim 1, characterized in that, In S1, the axes of the synchronously rotating coordinate system are estimated. Inject a high-frequency square wave voltage signal to extract the high-frequency current envelope containing position error. The specific content includes: estimating the synchronous rotating coordinate system The expression for injecting a high-frequency square wave voltage signal into the shaft is: ;in, ;in, To inject amplitude, For the injection signal period, The number of injection signal cycles, for High-frequency voltage components of the shaft. for High-frequency voltage components of the shaft. To inject a high-frequency voltage signal, The high-frequency voltage injection time point is used; the FOC algorithm is used to drive the motor rotation, and the three-phase current of the motor is obtained through current acquisition and Clarke transformation. The current values ​​of the two axes are used to extract the high-frequency current envelope containing the position error using a demodulation algorithm. The expression for the demodulation algorithm is: ;in, This indicates the estimated rotor electrical angle. They represent the extracted High-frequency current components of the shaft, This represents the inductance values ​​along the d-axis and q-axis. Indicates the electrical angle error value. This represents the switching factor corresponding to the polarity of the high-frequency square wave injection. An integer index that increments over time. The sampling period.

3. The sensorless control method for a permanent magnet synchronous motor across the entire speed range according to claim 2, characterized in that, The specific content of constructing a third-order nonlinear extended state observer in S2 includes: modeling the error tracking system PLL as a disturbed second-order system, and defining the state variables of the second-order system: ;in, For angular error, For speed error, This is an extended state, containing unmodeled dynamics and external disturbances. Indicates the angle error value. This represents the total external disturbance. Represent the derivative of the angle error; construct a discretized third-order nonlinear extended state observer, expressed as: ;in, System status Observed values; Input the error signal demodulated by HFI; The sampling period; The observer gain determines the observation bandwidth; It is the hyperbolic tangent function. To adjust the steepness of the nonlinear function, The observation error at each sampling point, Indicates the sampling point number; for the calculated state The predicted rotational speed is obtained by performing one integration, and the predicted angle is obtained by performing another integration.

4. The sensorless control method for a permanent magnet synchronous motor across the entire speed range according to claim 3, characterized in that, The specific content of building the improved sliding mode observer (SMO) model in S3 includes: defining a two-phase stationary coordinate system. The current state equation of the PMSM is as follows: Constructing a sliding mode observer: The sliding mode control signal is as follows: ;in, Indicates sliding mode gain. Represents a symbolic function. These represent the actual current value and the estimated current value, respectively. For motor phase resistance, For motor inductance, For motor The magnitude of the voltage on the shaft, This is the actual back electromotive force value of the motor. Predict the back electromotive force value for the motor; Includes back EMF information and high-frequency switching noise; processed using a second-order generalized integrator SOGI. The transfer function for SOGI is: ;in, Let be the transfer function of the second-order generalized integrator. This is the damping coefficient, used to adjust the SOGI's filter bandwidth and the system's dynamic response speed. For the Laplace operator; the center frequency of SOGI The estimated electric angular velocity is set in real time as the final overall solution output. Output after SOGI processing This is the fundamental back electromotive force.

5. The sensorless control method for a permanent magnet synchronous motor across the entire speed range according to claim 4, characterized in that, The position error in a back EMF phase-locked loop is defined as: ;in, These represent the fundamental back electromotive force values, This represents the electrical angle value predicted by the SMO. The position error signal is predicted by the SMO; this error is passed through a PI controller to output the rotational speed, and the angle is obtained by integration. 。 6. The sensorless control method for a permanent magnet synchronous motor across the entire speed range according to claim 5, characterized in that, The specific content of building a soft switching strategy based on hysteresis logic and vector synthesis in S5 includes: S501, constructing a hysteresis state mechanism based on speed feedback to improve the stability of the transition zone, and outputting the state interval to which the current speed belongs based on the hysteresis state mechanism; S502, calculating the fusion weight coefficient G based on the output of the hysteresis state mechanism and the current speed to form a smooth transition slope between the low-speed and high-speed zones; S503, using vector synthesis for angle fusion, and utilizing the four-quadrant arctangent function to synthesize the vector. The final fusion angle is obtained by performing calculations. The final output speed is obtained by calculating using a conventional linear weighted formula. 。 7. The sensorless control method for a permanent magnet synchronous motor across the entire speed range according to claim 6, characterized in that, The specific content of the hysteresis state mechanism based on speed feedback in S501 includes: preset three key speed thresholds, and low-speed cutoff speed. Switch to high-speed mode speed and the speed when switching to high-speed mode And satisfy Define a persistent state variable to indicate whether the current state is in pure high-speed mode. The specific logic judgment process is as follows: If the current state is in pure high-speed mode, the system will check if the absolute value of the feedback speed is reached. Fall to The following condition will trigger a switch to false, exiting pure high-speed mode; conversely, if currently in hybrid / low-speed mode, the state will be switched to false only when the absolute value of the feedback speed is... Exceed When the state is set to true, the system enters pure high-speed mode.

8. The sensorless control method for a permanent magnet synchronous motor across the entire speed range according to claim 7, characterized in that, The specific content of S502 includes: among which, the fusion weight coefficient The value ranges from 0 to 1, where 0 represents using HFI data exclusively and 1 represents using SMO data exclusively. When the state machine indicates that it is in pure high-speed mode, the weighting coefficient is forcibly adjusted. Set to 1.0 to directly transmit high-speed observation data to save computing resources; when the state machine indicates that it is in non-high-speed mode, perform linear interpolation calculation based on the real-time rotational speed: if the absolute value of the current rotational speed is less than or equal to... ,but Set to 0.0; if the absolute value of the current speed is within... and Between, then weight According to the formula Perform the calculation.

9. The sensorless control method for a permanent magnet synchronous motor across the entire speed range according to claim 8, characterized in that, The specific content of angle fusion using the S503 vector synthesis method includes: using weights to convert low-speed angles... With high speed angle Map each component to a weighted vector component on the unit circle to construct the x and y coordinates of the composite vector: Where G is the fusion weight coefficient; the synthesized vector is synthesized using the four-quadrant arctangent function. The final fusion angle is obtained by performing calculations. 。 10. A sensorless control method for a permanent magnet synchronous motor across the entire speed range according to claim 9, characterized in that, The expression for the conventional linear weighted formula is: ;in, This represents the rotational speed value calculated by the high-frequency injection method in the low-speed domain. This represents the rotational speed value calculated by the high-speed domain sliding mode observer method. Indicates the fusion weight coefficient. This represents the output rotational speed value after weighting.

Images