A PMSM control method based on an improved adaptive observer using LC filtering

By using the pole placement method and speed adaptive law of the improved adaptive full-order observer, the problems of complex observer design and insufficient estimation accuracy in LC-filtered PMSM drive systems are solved, and efficient online estimation of motor state variables and improvement of dynamic performance are achieved.

CN123122347APending Publication Date: 2026-11-17HARBIN INST OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611205066.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-10
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

The traditional sensorless control method for existing LC-filtered PMSM drive systems suffers from high observer order, large feedback gain matrix dimension, numerous parameters and complex debugging, and difficulty in maintaining consistent operating characteristics under different operating conditions, affecting the dynamic performance and estimation accuracy of the motor.

Method used

An improved adaptive full-order observer is adopted, and a pole placement method is designed through a simplified parameter tuning process. Combined with the speed adaptive law, online estimation of motor state variables is achieved, simplifying the parameter tuning process and maintaining consistent estimation performance over a wide speed range.

Benefits of technology

Without increasing hardware costs, accurate estimation of motor rotor position and speed was achieved, the resonance effect introduced by the filter was suppressed, and the dynamic performance and disturbance rejection capability of the system were improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122347A_ABST
    Figure CN123122347A_ABST
Patent Text Reader

Abstract

A control method for a permanent magnet synchronous motor (PMSM) based on an improved adaptive observer using an LC filter is disclosed, belonging to the field of permanent magnet synchronous motor control technology. The method comprises the following steps: Step 1: Sample the A and B phase currents of the inverter. The C phase current is obtained from the zero sum of the three-phase currents in a star connection, and the αβ axis current component is obtained through Clark transformation. Step 2: The MAFO receives the current and voltage components, designs the gain matrix based on a simplified parameter tuning process, and estimates the stator current and rotor position online. Step 3: The stator current, inverter output current, and rotor position are transformed using Park transformation to obtain the dq axis stator current and inverter output current. Step 4: The dq axis stator current is used for inner-loop feedback and PI regulation. The difference between the dq axis inverter output and stator current is used as the capacitor current, and active damping negative feedback is fed to the PI output to suppress resonance, obtaining the dq voltage reference. Step 5: The dq axis voltage reference is inversely transformed using Park transformation to obtain the αβ voltage, which is sent to SVPWM to generate a drive signal and fed back to the MAFO. This invention is applicable to sensorless control of LC-filtered three-phase permanent magnet synchronous motor drive systems.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of permanent magnet synchronous motor control technology, specifically relating to an LC-filtered PMSM control method based on an improved adaptive observer. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) have been widely used in various voltage levels and power ranges of drive applications due to their excellent efficiency, high power density, and relatively simple control structure. These applications include mine hoisting equipment, rail transit traction, ship electric propulsion, and high-end servo systems. Under normal circumstances, PMSMs are directly powered by voltage source inverters, eliminating the need for output filtering. However, in extreme deep-ground environments such as mines and oil and gas extraction sites, the inverter and motor are often far apart, connected by long-distance power cables. In such cases, the distributed inductance and parasitic capacitance of the cable can cause voltage traveling wave reflection, leading to overvoltage surges and high-frequency damped oscillations at the motor stator. This severely threatens the lifespan of the stator winding insulation. If the overvoltage amplitude exceeds the tolerance limit, it can cause insulation breakdown, resulting in significant economic losses.

[0003] To effectively alleviate the aforementioned problems, an LC filter is often inserted in series at the inverter output port in engineering to filter out high-order harmonic components in the pulse-width modulated voltage, making the terminal voltage waveform supplied to the motor more sinusoidal, thereby significantly reducing the voltage reflection effect caused by long-line transmission and its destructive impact on the insulation system. However, the introduction of the LC filter transforms the load driven by the inverter into a high-order, nonlinear, and strongly coupled dynamic system composed of the filter network and the PMSM. Under this topology, if the traditional sensorless control strategy is used, additional voltage and current sensing elements are usually required to obtain the state variable information of the motor itself. This not only disrupts the original hardware structure of the drive system but also leads to increased costs and decreased reliability.

[0004] Therefore, researching a high-performance sensorless control scheme suitable for LC-filtered PMSM drive systems while maintaining existing hardware configurations is a pressing engineering need. Currently, existing engineering applications have constructed full-order state observers in a two-phase stationary coordinate system to achieve online estimation of the motor system's state variables and rotor position. However, these observers have high order, large feedback gain matrix dimensions, and require numerous parameters to be tuned, resulting in complex debugging procedures and significantly limiting their convenience and versatility in engineering practice. Furthermore, these observers struggle to maintain consistent operating characteristics under different motor operating conditions, which significantly reduces the motor's wide-speed-range performance, and the slow convergence of the estimation process negatively impacts the motor's dynamic performance. Summary of the Invention

[0005] The purpose of this invention is to solve the above-mentioned problems existing in the prior art, and to provide an LC-filtered PMSM control method based on an improved adaptive observer.

[0006] This invention designs an improved adaptive full-order observer, whose key feature is that the pole placement of the improved observer can be completed through a very simple parameter tuning process, thereby significantly reducing the workload of observer design and parameter tuning. Simultaneously, the improved speed adaptive law makes the parameter setting of the speed estimation stage extremely simple and accurate, and the estimated performance of the designed observer remains consistent under different motor operating conditions. Furthermore, the convergence process of the estimation is significantly improved, enhancing the dynamic performance of the motor. Experimental results fully demonstrate that the improved adaptive full-order observer proposed in this invention can maintain high estimation accuracy, good disturbance rejection capability, and excellent dynamic performance over a wide speed range of motor operation.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] A PMSM control method based on an improved adaptive observer using LC filtering includes the following steps:

[0009] Step 1, Signal Sampling and Stationary Coordinate Transformation: The current signals of phases A and B at the inverter output are obtained by sampling with a current sensor, and the current signal of phase C is derived based on the fact that the sum of the instantaneous values ​​of the three-phase currents in the star-connected PMSM is zero; then, the current components of the inverter output under the αβ axis in the two-phase stationary coordinate system are obtained by Clark transformation, and used as the input variables for subsequent MAFO, so as to realize the online real-time estimation of MAFO variables.

[0010] Step 2, MAFO Online Estimation: The inverter output current component under the αβ axis of the two-phase stationary coordinate system obtained in Step 1 and the inverter output voltage component under the αβ axis of the current inner loop PI control are jointly input into the MAFO calculation model; the MAFO is designed with a feedback gain matrix based on the pole placement method of a simplified parameter tuning process to achieve online estimation of the MAFO state variables; the online estimation includes the PMSM stator current signal under the αβ axis and the PMSM rotor position information estimated by MAFO;

[0011] Step 3, Rotating Coordinate Transformation: The PMSM stator current signal under the αβ axis of the two-phase stationary coordinate system estimated by MAFO in Step 2, the inverter output current component under the αβ axis calculated in Step 1, and the PMSM rotor position information estimated by MAFO in Step 2 are used as input variables for the rotating coordinate transformation stage; through Park transformation, the PMSM stator current signal under the αβ axis and the inverter output current component are uniformly converted into current components under the dq axis of the two-phase rotating coordinate system.

[0012] Step 4, Inner Current Loop Control: The estimated PMSM stator current on the dq axis obtained after the rotation coordinate transformation in Step 3 is used as the closed-loop feedback variable of the inner current loop, and is adjusted by the inner current loop PI controller. At the same time, the difference between the estimated inverter output current on the dq axis and the estimated PMSM stator current on the dq axis is taken as the dq axis capacitor current, which is superimposed on the output of the inner current loop PI controller in the form of negative feedback to ensure stable system operation. The inner current loop finally outputs the inverter output voltage reference value on the dq axis, which is used by subsequent links to control the switching action of the inverter.

[0013] Step 5, Signal Conversion: The inverter output voltage reference value under the dq axis output from the current inner loop control loop in Step 4 is transformed by inverse Park to obtain the voltage signal under the two-phase stationary αβ coordinate system. On the one hand, it is sent to the SVPWM module to generate a three-phase chopper drive signal to drive the PMSM control. On the other hand, it is fed back to MAFO as the input variable of MAFO for variable iteration estimation in the next control cycle.

[0014] Furthermore, in step 1, the current signals of phases A and B at the inverter output are obtained by sampling with a current sensor, and the current signal of phase C is derived based on the fact that the sum of the instantaneous values ​​of the three-phase currents in the star-connected PMSM is zero; then, the inverter output current components in the two-phase stationary coordinate system αβ axis are obtained through Clark transformation; the calculation formula is as follows:

[0015] (1)

[0016] (2)

[0017] in, , , This represents the inverter output current in a three-phase stationary coordinate system. , This represents the inverter output current in a two-phase stationary coordinate system.

[0018] Furthermore, in step 2, the inverter output current component under the αβ axis of the two-phase stationary coordinate system obtained in step 1 and the inverter output voltage component under the αβ axis of the current inner loop PI control are jointly input into the calculation model of MAFO; the MAFO is designed based on the pole placement method of a simplified parameter tuning process to design the feedback gain matrix, so as to realize the online estimation of MAFO state variables; its calculation formula is as follows:

[0019] (3)

[0020] In the formula

[0021]

[0022]

[0023]

[0024] (4)

[0025] in, ; , , and These represent the inverter output current, PMSM stator voltage, PMSM stator current, and PMSM extended back electromotive force, respectively. It is the output voltage of the inverter; , , and These represent the LC filter inductance, LC filter capacitance, PMSM stator resistance, and... Shaft inductance; The estimated rotor electric angular frequency of the PMSM; These are the low-pass filter coefficients; For the Laplace operator; ; To iteratively optimize the step size; Estimate rotor position for PMSM; This is the feedback gain matrix; The sampling period; This indicates the current control cycle.

[0026] Furthermore, in step 3, the PMSM stator current signal in the two-phase stationary coordinate system αβ axis estimated by MAFO in step 2, the inverter output current component in the αβ axis calculated in step 1, and the PMSM rotor position information estimated by MAFO in step 2 are used together as input variables for the rotating coordinate transformation stage. By performing the Park transformation, the PMSM stator current signal in the αβ axis and the inverter output current component are uniformly converted into current components in the two-phase rotating coordinate system dq axis; the calculation formula is as follows:

[0027] (5)

[0028] (6)

[0029] in, , , , These represent the d-axis estimated inverter output current, the q-axis estimated inverter output current, the d-axis estimated PMSM stator current, and the q-axis estimated PMSM stator current, respectively. , These represent the α-axis estimated PMSM stator current and the β-axis estimated PMSM stator current, respectively.

[0030] Furthermore, in step 4, the inner current loop ultimately outputs the inverter output voltage reference value along the dq axis; its calculation formula is as follows:

[0031] (7)

[0032] in, , This represents the proportional coefficient of the dq-axis current inner loop PI controller. , This represents the integral coefficient of the dq-axis PI controller within the current inner loop. Represents the Laplace operator; , express Reference value of inverter output voltage under the dq axis at any given time; , This indicates the reference value of the PMSM stator current under the dq axis; and These represent the active damping coefficients along the d and q axes, respectively. This represents the delay operator per unit sampling period.

[0033] Furthermore, in step 5, the voltage signal in the two-phase stationary αβ coordinate system is obtained through inverse Park transform; its calculation formula is as follows:

[0034] (8)

[0035] in, , This represents the voltage signal in a two-phase stationary αβ coordinate system.

[0036] Furthermore, in step 2, to improve the convergence speed of the estimation process and enhance the system's response capability under dynamic operating conditions, the estimated PMSM extended back EMF and the extracted corresponding variable estimation error are combined. Together, they serve as inputs to the speed adaptive law; the speed estimation error correction term is pre-added to the speed estimation stage through a feedforward method to accelerate the convergence process of the adaptive law, thereby accurately estimating the PMSM rotor position information.

[0037] The advantages of this invention over the prior art are:

[0038] Traditional sensorless control methods for LC-filtered permanent magnet synchronous motor (PMSM) drive systems typically rely on full-order state observers with high-dimensional feedback gain matrices. The high order of the observer results in a large-dimensional feedback gain matrix, numerous adjustable parameters, and a complex tuning process. Furthermore, sliding mode control or adaptive strategies introduced in some schemes to improve stability further increase algorithm complexity and computational burden. Some methods even introduce new instabilities due to pole shifts during wide-speed operation, negatively impacting the system's control performance. Moreover, existing estimation methods have slow convergence speeds, reducing the motor's dynamic performance. To address these issues, many projects employ low-order observer schemes that only include the motor model to reconstruct motor state variables. However, this requires additional voltage or current sensors, disrupting the original hardware structure, increasing system cost, and reducing operational reliability. Therefore, for LC-filtered PMSM systems, traditional observer designs present a trade-off between system estimation accuracy, dynamic response speed, and ease of parameter tuning.

[0039] This invention proposes an LC-filtered PMSM control method based on an improved adaptive observer. This method does not alter the original main circuit topology and control structure of the LC-filtered permanent magnet synchronous motor drive system. It only requires executing an optimization program in a digital signal processor to estimate the controlled object's state variables and motor speed, without incurring additional hardware costs or energy losses from electronic components. The operation, control, use, and maintenance methods remain consistent with the original system. When applied to an LC-filtered permanent magnet synchronous motor drive system, it can achieve accurate rotor position and speed estimation over a wide speed range without adding additional voltage or current sensors. It effectively suppresses the resonance effects introduced by the filter, enabling the motor drive system to reach the given desired value more quickly and smoothly. This method has significant engineering application prospects in long cable drives such as mining equipment, rail transportation, ship propulsion, and high-precision servo systems.

[0040] This invention is applicable to sensorless control of LC-filtered three-phase permanent magnet synchronous motor (PMSM) drive systems. This control method aims to solve the problems of design complexity and parameter tuning difficulties caused by the high-dimensional gain matrix of traditional full-order observers. By utilizing a pole placement method based on a simplified parameter tuning process, combined with an improved speed adaptive law, it achieves accurate synchronous estimation of the motor's stator current, stator voltage, extended back EMF, rotor position, and speed. This reduces system debugging complexity while effectively suppressing the increase in estimated rotor position errors and oscillations caused by parameter mismatch and speed factors under wide-speed-range operating conditions. Attached Figure Description

[0041] Figure 1 This is a block diagram of the LC-filtered PMSM control method based on an improved adaptive observer according to the present invention.

[0042] Figure 2 This is a block diagram of an improved adaptive full-order observer.

[0043] Figure 3 This is a comparison chart of the estimated rotor position and speed error during PMSM acceleration and deceleration under load; where:

[0044] Figure 3 (a) is a diagram showing the error between rotor position and speed estimated using the method of the present invention during PMSM acceleration and deceleration under load;

[0045] Figure 3 (b) is a diagram showing the error in rotor position and speed estimated using traditional methods during PMSM acceleration and deceleration under load.

[0046] Figure 4 This is a comparison chart of the estimated rotor position and speed errors during sudden load changes in a PMSM (Power Management System); where:

[0047] Figure 4(a) is a diagram showing the error in rotor position and speed estimated using the method of the present invention when the PMSM load changes abruptly.

[0048] Figure 4 (b) is a diagram showing the error in rotor position and speed estimated using traditional methods when the PMSM load changes abruptly. Detailed Implementation

[0049] This invention uses a permanent magnet synchronous motor The vector control algorithm uses PI controllers for both the outer loop speed controller and the inner loop current controller. The PMSM rotor speed used for closed-loop feedback of system speed and the PMSM rotor position used for rotating coordinate transformation are both estimated online using the improved adaptive full-order observer (MAFO) described in this invention.

[0050] like Figure 1 As shown, from the perspective of the signal transmission sequence, in the LC-filtered PMSM control system, both the PMSM speed controller and the current inner loop controller use PI controllers to regulate the PMSM speed and stator current. Variables with the dq sign represent components in the rotating coordinate system. The variables in the symbols represent the components in the two-phase stationary coordinate system. PMSM represents a permanent magnet synchronous motor. An LC filter is installed between the inverter and the PMSM to attenuate harmonics in the inverter output voltage and reduce the output voltage variability. SVPWM represents space vector modulation. DC bus voltage, and These represent the actual PMSM rotor position and speed information, respectively. and These represent the PMSM rotor position and speed information estimated by the method described in this invention, respectively.

[0051] like Figure 1 As shown, this embodiment provides an LC-filtered PMSM control method based on an improved adaptive observer, which includes the following steps:

[0052] Step 1: Signal Sampling and Stationary Coordinate Transformation

[0053] First, the current signals of phases A and B at the inverter output are sampled using an overcurrent sensor. Then, based on Kirchhoff's current law (that the sum of the instantaneous values ​​of the three-phase currents in a star-connected PMSM is zero), the current signal of phase C is calculated. Next, a Clark transform is performed on the acquired three-phase current signals to obtain the inverter output current components in the two-phase stationary coordinate system αβ axis. The calculation formula is as follows:

[0054] (1)

[0055] (2)

[0056] in, , , This represents the inverter output current in a three-phase stationary rotating coordinate system. , The inverter output current is represented in a two-phase stationary coordinate system; and the inverter output current component is used as the input variable of the subsequent MAFO (Modified Adaptive Full-Order Observer) to achieve online real-time estimation of MAFO variables.

[0057] Step 2: Online estimation of MAFO

[0058] The inverter output current component under the αβ axis of the two-phase stationary coordinate system obtained in step 1, and the inverter output voltage component under the αβ axis of the current inner loop PI control output are jointly input into the calculation model of MAFO (Improved Adaptive Full-Order Observer) proposed in this invention. The MAFO uses the pole placement method based on a simplified parameter tuning process to design the feedback gain matrix, so as to realize the online estimation of the MAFO state variables. Its calculation formula is as follows:

[0059] (3)

[0060] In the formula

[0061]

[0062]

[0063]

[0064] (4)

[0065] in, ; , , and These represent the inverter output current, PMSM stator voltage, PMSM stator current, and PMSM extended back electromotive force, respectively. It is the output voltage of the inverter; , , and These represent the LC filter inductance, LC filter capacitance, PMSM stator resistance, and... Shaft inductance; The estimated rotor electric angular frequency of the PMSM; These are the low-pass filter coefficients; For the Laplace operator; ; To iteratively optimize the step size; Estimate rotor position for PMSM; This is the feedback gain matrix; The sampling period; This indicates the current control cycle.

[0066] The online estimation includes the motor stator current signal under the αβ axis and the estimated motor rotor position.

[0067] Equation (3) is the computational model expression of the improved adaptive full-order observer; Equation (4) is the feedback gain matrix, which is part of the computational model of the original improved adaptive full-order observer.

[0068] Furthermore, in step 2, to improve the convergence speed of the estimation process and enhance the system's response capability under dynamic operating conditions, the estimated PMSM extended back EMF and the extracted corresponding variable estimation error are combined. Together, they serve as inputs to the speed adaptive law; the speed estimation error correction term is pre-added to the speed estimation stage via a feedforward method to accelerate the convergence process of the adaptive law, thereby accurately estimating the PMSM rotor position information (see equation (3)). This ensures that the observer maintains excellent estimation performance across the entire speed range. The estimated PMSM stator current signal in the two-phase stationary coordinate system αβ axis and the estimated PMSM rotor position information are used as input variables for the rotating coordinate transformation stage to achieve system variable transformation to the rotating coordinate system.

[0069] Step 3: Rotation coordinate transformation

[0070] The PMSM stator current signal in the two-phase stationary coordinate system αβ axis estimated by MAFO in step 2, the inverter output current component in the αβ axis calculated in step 1, and the PMSM rotor position information estimated by MAFO in step 2 are used as input variables for the rotating coordinate transformation stage. Through Park transformation, the PMSM stator current signal in the αβ axis and the inverter output current component are uniformly converted into current components in the two-phase rotating coordinate system dq axis. The dq axis estimated PMSM stator current and dq axis estimated inverter output current output after transformation are used as inputs for the subsequent current inner loop control stage to achieve closed-loop control of the PMSM stator current.

[0071] Step 4: Inner current loop control

[0072] The estimated PMSM stator current obtained after the rotation coordinate transformation in step 3 is used as the closed-loop feedback variable of the inner current loop. The PMSM stator current is regulated and controlled by the inner current loop PI controller. At the same time, the difference between the estimated inverter output current and the estimated PMSM stator current is taken as the dq axis capacitor current and used as a negative feedback variable. This difference is superimposed on the output of the inner current loop PI controller to ensure stable system operation. The inner current loop finally outputs the inverter output voltage reference value under the dq axis and uses it as the input of the subsequent conversion stage to control the switching action of the inverter.

[0073] The inner current loop ultimately outputs the inverter output voltage reference value along the dq axis; its calculation formula is as follows:

[0074] (7)

[0075] in, , This represents the proportional coefficient of the dq-axis current inner loop PI controller. , This represents the integral coefficient of the dq-axis PI controller within the current inner loop. Represents the Laplace operator; , express Reference value of inverter output voltage under the dq axis at any given time; , This indicates the reference value of the PMSM current of the motor under the dq axis; and These represent the active damping coefficients along the d and q axes, respectively. This represents the delay operator per unit sampling period.

[0076] Step 5: Signal Conversion

[0077] The inverter output voltage reference value under the dq axis output from the current inner loop control loop in step 4 is transformed by inverse Park to obtain the voltage signal under the two-phase stationary αβ coordinate system. On the one hand, it is sent to the SVPWM (Space Vector Pulse Width Modulation) module to generate a three-phase chopper drive signal to drive the PMSM control. On the other hand, it is fed back to MAFO (Improved Adaptive Full-Order Observer) as the input variable of MAFO for variable iterative estimation in the next control cycle.

[0078] The voltage signal, converted to a two-phase stationary coordinate system by the inverse Park transform, is calculated using the following formula:

[0079] (8)

[0080] in, , This represents the voltage signal in a two-phase stationary αβ coordinate system.

[0081] This invention proposes an LC-filtered PMSM control method based on an improved adaptive observer, such as... Figure 1 As shown in the figure. Based on the mathematical model of the LC-filtered permanent magnet synchronous motor drive system, a full-order state observer model including filter inductor current, stator voltage, stator current, and motor extended back EMF is established. The observer pole placement method is derived, and the optimal pole locations and feedback gain matrix required by the observer under different operating conditions are calculated. Furthermore, considering practicality and convenience, this invention adopts a pole placement method based on a simplified parameter tuning process to complete the design of the feedback gain matrix. By setting appropriate observer pole locations, unstable regions are dynamically avoided, thereby significantly reducing the complexity of parameter tuning while ensuring estimation accuracy. The corresponding logic block diagram is shown in the figure. Figure 2 As shown in the experimental results. Figure 3 (a) It can be seen that the sensorless drive system for permanent magnet synchronous motors based on the proposed improved adaptive full-order observer exhibits a maximum speed estimation error of only 0.2 rad / s and a maximum position estimation error of only 4.5° during PMSM loading and deceleration. This effectively suppresses the impact of LC filter resonance on system stability while ensuring high-precision state estimation over a wide speed range. Figure 3 (b) The traditional method shown has a maximum velocity estimation error of 8 rad / s and a maximum position estimation error of 8.9° under the same operating conditions. From the experimental results... Figure 4 (a) It can be seen that when the PMSM operates at a frequency of 420 rad / s and is subjected to a step load disturbance, the proposed method has a velocity estimation error of 27.3 rad / s and a position estimation error of 12.1°, which is significantly lower than... Figure 4 (b) The conventional method resulted in a reduction of 28.5% and 26.2%, respectively. Therefore, the method proposed in this invention significantly improves the system's disturbance rejection capability and estimation robustness while ensuring dynamic response performance.

Claims

1. A PMSM control method based on an improved adaptive observer using LC filtering, characterized in that: Includes the following steps: Step 1, Signal Sampling and Stationary Coordinate Transformation: The current signals of phases A and B at the inverter output are obtained by sampling with a current sensor, and the current signal of phase C is derived based on the fact that the sum of the instantaneous values ​​of the three-phase currents in the star-connected PMSM is zero; then, the current components of the inverter output under the αβ axis in the two-phase stationary coordinate system are obtained by Clark transformation, and used as the input variables for subsequent MAFO, so as to realize the online real-time estimation of MAFO variables. Step 2, MAFO Online Estimation: The inverter output current component under the αβ axis of the two-phase stationary coordinate system obtained in Step 1 and the inverter output voltage component under the αβ axis of the current inner loop PI control are jointly input into the MAFO calculation model; the MAFO is designed with a feedback gain matrix based on the pole placement method of a simplified parameter tuning process to achieve online estimation of the MAFO state variables; the online estimation includes the PMSM stator current signal under the αβ axis and the PMSM rotor position information estimated by MAFO; Step 3, Rotating Coordinate Transformation: The PMSM stator current signal under the αβ axis of the two-phase stationary coordinate system estimated by MAFO in Step 2, the inverter output current component under the αβ axis calculated in Step 1, and the PMSM rotor position information estimated by MAFO in Step 2 are used as input variables for the rotating coordinate transformation stage; through Park transformation, the PMSM stator current signal under the αβ axis and the inverter output current component are uniformly converted into current components under the dq axis of the two-phase rotating coordinate system. Step 4, Inner Current Loop Control: The estimated PMSM stator current on the dq axis obtained after the rotation coordinate transformation in Step 3 is used as the closed-loop feedback variable of the inner current loop, and is adjusted by the inner current loop PI controller. At the same time, the difference between the estimated inverter output current on the dq axis and the estimated PMSM stator current on the dq axis is taken as the dq axis capacitor current, which is superimposed on the output of the inner current loop PI controller in the form of negative feedback to ensure stable system operation. The inner current loop finally outputs the inverter output voltage reference value on the dq axis, which is used by subsequent links to control the switching action of the inverter. Step 5, Signal Conversion: The inverter output voltage reference value under the dq axis output from the current inner loop control loop in Step 4 is transformed by inverse Park to obtain the voltage signal under the two-phase stationary αβ coordinate system. On the one hand, it is sent to the SVPWM module to generate a three-phase chopper drive signal to drive the PMSM control. On the other hand, it is fed back to MAFO as the input variable of MAFO for variable iteration estimation in the next control cycle.

2. The LC-filtered PMSM control method according to claim 1, characterized in that: In step 1, the current signals of phases A and B at the inverter output are obtained by sampling with a current sensor, and the current signal of phase C is derived based on the fact that the sum of the instantaneous values ​​of the three-phase currents in the star-connected PMSM is zero; then, the inverter output current components in the two-phase stationary coordinate system αβ axis are obtained through Clark transformation; the calculation formula is as follows: (1) (2) in, , , This represents the inverter output current in a three-phase stationary coordinate system. , This represents the inverter output current in a two-phase stationary coordinate system.

3. The LC-filtered PMSM control method according to claim 2, characterized in that: In step 2, the inverter output current component under the αβ axis of the two-phase stationary coordinate system obtained in step 1 and the inverter output voltage component under the αβ axis of the current inner loop PI control are jointly input into the calculation model of MAFO. The MAFO is designed based on the pole placement method of a simplified parameter tuning process to design the feedback gain matrix, so as to realize the online estimation of MAFO state variables. Its calculation formula is as follows: (3) In the formula (4) in, ; , , and These represent the inverter output current, PMSM stator voltage, PMSM stator current, and PMSM extended back electromotive force, respectively. It is the output voltage of the inverter; , , and These represent the LC filter inductance, LC filter capacitance, PMSM stator resistance, and... Shaft inductance; The estimated rotor electric angular frequency of the PMSM; These are the low-pass filter coefficients; For the Laplace operator; ; To iteratively optimize the step size; Estimate rotor position for PMSM; This is the feedback gain matrix; The sampling period; This indicates the current control cycle.

4. The LC-filtered PMSM control method according to claim 3, characterized in that: In step 3, the PMSM stator current signal in the two-phase stationary coordinate system αβ axis estimated by MAFO in step 2, the inverter output current component in the αβ axis calculated in step 1, and the PMSM rotor position information estimated by MAFO in step 2 are used as input variables for the rotating coordinate transformation stage. By performing the Park transformation, the PMSM stator current signal in the αβ axis and the inverter output current component are uniformly converted into current components in the two-phase rotating coordinate system dq axis. The calculation formula is as follows: (5) (6) in, , , , These represent the d-axis estimated inverter output current, the q-axis estimated inverter output current, the d-axis estimated PMSM stator current, and the q-axis estimated PMSM stator current, respectively. , These represent the α-axis estimated PMSM stator current and the β-axis estimated PMSM stator current, respectively.

5. The LC-filtered PMSM control method according to claim 4, characterized in that: In step 4, the inner current loop ultimately outputs the inverter output voltage reference value along the dq axis; its calculation formula is as follows: (7) in, , This represents the proportional coefficient of the dq-axis current inner loop PI controller. , This represents the integral coefficient of the dq-axis PI controller within the current inner loop. Represents the Laplace operator; , express Reference value of inverter output voltage under the dq axis at any given time; , This indicates the reference value of the PMSM stator current under the dq axis; and These represent the active damping coefficients along the d and q axes, respectively. This represents the delay operator per unit sampling period.

6. The LC-filtered PMSM control method according to claim 5, characterized in that: In step 5, the voltage signal in the two-phase stationary αβ coordinate system is obtained through inverse Park transform; its calculation formula is as follows: (8) in, , This represents the voltage signal in a two-phase stationary αβ coordinate system.

7. The LC-filtered PMSM control method according to claim 1, characterized in that: In step 2, to improve the convergence speed of the estimation process and enhance the system's response under dynamic conditions, the estimated PMSM extended back EMF and the extracted corresponding variable estimation error are combined. Together they serve as inputs to the velocity adaptive law; The speed estimation error correction term is pre-added to the speed estimation stage through a feedforward method to accelerate the convergence process of the adaptive law, thereby accurately estimating the PMSM rotor position information.