Model reference adaptive based on-line parameter identification method for surface-mounted permanent magnet synchronous motor

Abstract

The application particularly relates to a model reference adaptive based surface-mounted permanent magnet synchronous motor parameter online identification method, which comprises the following steps: constructing a d-q coordinate system adjustable mathematical model, collecting voltage, current and rotating speed signals, solving current estimated values and generating double-axis current errors; constructing a parallel adaptive update law based on the errors, synchronously updating parameters to be identified, and combining double-axis threshold values and multi-period persistence criteria to complete convergence judgment. The method does not need to stop and inject high-frequency excitation, can synchronously identify stator resistance, stator synchronous inductance and rotor permanent magnet flux linkage under a loaded working condition, track parameter drift caused by temperature rise and magnetic saturation, improve multi-parameter identification decoupling capability and convergence stability, has low operation cost, is convenient to realize on a general digital operation chip, and is suitable for motor online parameter correction.

Description

Technical Field

[0001] This invention relates to the field of permanent magnet synchronous motor technology, and specifically to an online parameter identification method for surface-mounted permanent magnet synchronous motors based on model reference adaptation. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs), with their high power density, high operating efficiency, high torque-to-inertia ratio, and excellent dynamic response characteristics, are widely used as core drive components in high-end equipment fields such as electric vehicles, industrial robots, aerospace, and CNC machine tools. Their control performance directly determines the operating accuracy and energy efficiency of the entire system. In the high-performance control system of PMSMs, field-oriented control (FOC) is the current mainstream core control strategy. This strategy decouples the stator current into independent excitation components (d-axis current) and torque components (q-axis current) through coordinate transformation technology, achieving independent and precise control of motor torque and magnetic flux. This is the technical foundation for ensuring the high-performance operation of the motor.

[0003] However, the control accuracy and dynamic performance of field-oriented control are highly dependent on the accuracy of the motor's core electrical parameters (mainly stator resistance, stator inductance, and rotor permanent magnet flux linkage). During actual motor operation, dynamic changes in operating conditions cause these core electrical parameters to drift in real time: stator resistance increases with rising winding operating temperature, stator inductance decreases due to magnetic circuit saturation caused by increased load current, and rotor permanent magnet flux linkage undergoes irreversible slight demagnetization due to temperature increases. In other words, the time-varying characteristics of these electrical parameters directly lead to a decrease in the decoupling accuracy of field-oriented control, resulting in a series of problems such as reduced torque control accuracy, lag in system dynamic response, and deterioration of motor operating efficiency, severely restricting the motor's control performance across all operating conditions. Therefore, high-precision online identification of key electrical parameters of permanent magnet synchronous motors and real-time acquisition of actual parameter values ​​under different operating conditions are the core prerequisite and key support for achieving high-performance and stable motor control.

[0004] Currently, traditional technical approaches for online parameter identification of permanent magnet synchronous motors mainly include three categories: extended Kalman filtering, recursive least squares, and high-frequency signal injection. Among them, the extended Kalman filtering method can achieve synchronous estimation of multiple parameters and multiple state variables, possessing a certain degree of noise resistance. However, this algorithm has a large computational load, demanding stringent computing resources from the main control chip. Furthermore, the filtering parameters are difficult to tune, and its engineering adaptability is poor, making large-scale application in practical industrial scenarios difficult. While the recursive least squares method has a simple structure and is easy to implement, it is highly sensitive to current and voltage sampling noise, and is prone to data saturation during long-term continuous operation, leading to a significant decrease in the dynamic tracking capability of parameters and failing to meet the continuous identification requirements under wide operating conditions. The high-frequency signal injection method utilizes the salient pole characteristics and magnetic circuit saturation effect of the motor to complete parameter identification without relying on the motor's fundamental frequency model. However, it requires injecting additional high-frequency disturbance signals into the motor system. This method is highly susceptible to additional torque pulsation, operating noise, and power loss, disrupting the motor's operational stability and reducing the overall system efficiency, making it unsuitable for applications with high requirements for operational stability. It is evident that these three traditional identification methods all have inherent defects that are difficult to overcome. They cannot simultaneously balance identification accuracy, engineering feasibility, and system operation stability, and are therefore unable to support the full-range high-performance control requirements of permanent magnet synchronous motors.

[0005] To address the inherent drawbacks of traditional identification algorithms, such as high computational complexity, weak robustness, and susceptibility to intrusive disturbances, those skilled in the art have introduced the Model Reference Adaptive System (MRAS) method into the field of online parameter identification for permanent magnet synchronous motors. This method boasts a simple and clear structure, its stability can be rigorously proven theoretically, and it offers moderate computational complexity and strong engineering feasibility. However, in practical engineering applications, existing identification techniques based on MRAS still have many shortcomings and cannot meet the stringent requirements of high-performance control.

[0006] For example, the paper "A Sensorless Predictive Control Method Based on Parameter Identification of Permanent Magnet Synchronous Motor" (CN112350632B) achieves sensorless predictive control by building a reference model and an adaptive observation model in the α-β stationary coordinate system and combining parameter identification. However, this scheme only performs online identification of stator inductance and rotor flux linkage, which cannot fully cover the time-varying characteristics of the motor's core parameters and is difficult to solve the control performance degradation problem caused by stator resistance drift. The completeness and applicability of parameter identification are significantly limited.

[0007] For example, the paper "Online Parameter Identification Method for Surface-Mounted Permanent Magnet Synchronous Motor Based on Model Reference Adaptation" (CN112803854A) constructs an identification system by combining a finite set model predictive control framework, which optimizes the control architecture. However, it only identifies two parameters: stator inductance and rotor flux linkage. It cannot comprehensively and accurately perceive the real-time operating status of the motor, making it difficult to support the requirements of high-precision full-domain control.

[0008] To overcome the limitations of incomplete coverage of the aforementioned core parameters, those skilled in the art have further iterated on the technical solutions and proposed improvements for multi-parameter synchronous identification. For example, the publication CN118611498A, "Online Identification Method for Multi-parameters of Permanent Magnet Synchronous Motors Based on Model Reference Adaptation," constructs a model reference adaptive system for three-parameter synchronous identification, which can simultaneously achieve online identification of stator resistance, stator inductance, and rotor flux linkage. To achieve stable decoupled identification of multiple parameters, this solution requires actively switching specific operating modes such as d-axis current setting conditions and entering field weakening operation during the identification process to create conditions for parameter identification. However, in actual engineering verification, it was found that this type of solution did not fundamentally solve the problem but instead gave rise to new technical problems, as follows:

[0009] (1) The control logic for active operating condition switching significantly increases the complexity of the control algorithm, raising the debugging cost and hardware implementation threshold for engineering implementation;

[0010] (2) The switching process of operating conditions will cause intrusive interference to the normal and stable operation of the motor, which will disrupt the continuity and stability of the system output and make it difficult to adapt to application scenarios with strict requirements for smooth operation.

[0011] (3) The parameter identification real-time performance and dynamic tracking performance of this type of scheme still have obvious shortcomings. In scenarios with rapidly changing working conditions, it is very easy to encounter problems such as identification lag and tracking accuracy decay, which in turn restricts the further improvement of motor control performance.

[0012] In summary, although technicians have conducted continuous research and improvement on the online parameter identification technology for permanent magnet synchronous motors, there are still many irreconcilable technical defects, such as incomplete parameter identification, difficulty in algorithm engineering implementation, strong disturbance intrusion, and dynamic response lag, which cannot meet the needs of high-end equipment fields for high-performance and stable control of permanent magnet synchronous motors across the entire domain. Summary of the Invention

[0013] The purpose of this invention is to address the shortcomings of existing technologies by providing a model reference adaptive online parameter identification method for surface-mounted permanent magnet synchronous motors. This method eliminates the need for motor shutdown and additional high-frequency excitation signals, and can simultaneously identify three key electrical parameters—stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage—under normal motor operation under load. It can also track parameter drift caused by winding temperature rise and magnetic saturation effects in real time. Furthermore, it improves the decoupling capability and convergence stability of the multi-parameter synchronous identification process, reduces algorithm computational overhead, and facilitates engineering implementation on general-purpose digital computing chips.

[0014] The objective of this invention is achieved through the following approach:

[0015] A model-reference adaptive online parameter identification method for surface-mounted permanent magnet synchronous motors, executed cyclically according to the control cycle of the field-oriented control system, includes the following steps:

[0016] 1) Based on the premise that the d-axis inductance and the q-axis inductance are equal, construct an adjustable mathematical model of the permanent magnet synchronous motor in the dq rotating coordinate system containing the set of parameters to be identified. The set of parameters to be identified includes the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage.

[0017] 2) During each control cycle of normal motor operation, the stator voltage component, stator current component, and rotor electric angular velocity in the dq rotating coordinate system are collected in real time. The normal operation process of the motor is the d-axis stator current. Furthermore, the steady-state operation of the field-oriented control system is characterized by no drastic changes in motor speed and load torque.

[0018] 3) Using the stator voltage component, stator current component, rotor electric angular velocity, and the parameter set to be identified updated in the previous control cycle obtained in step 2), input the adjustable mathematical model of the permanent magnet synchronous motor to calculate the estimated value of the stator current.

[0019] 4) The difference between the real-time acquired stator current components and the estimated stator current is calculated to generate a current error signal;

[0020] 5) Construct a parallel adaptive update law. The parallel adaptive update law takes the stator current component and rotor electric angular velocity obtained in real time in step 2) and the current error signal obtained in step 4) as inputs. Under the premise of not changing the current operating conditions of the motor, it synchronously updates the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage.

[0021] 6) Determine whether the current error signal has converged. If it has converged, then output the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage obtained by parallel adaptive update law iteration as the actual electrical parameters of the motor under the current operating condition.

[0022] Preferably, in step 1), while constructing the adjustable mathematical model of the permanent magnet synchronous motor, the set of parameters to be identified is initialized. The initial values ​​of the set of parameters to be identified are assigned and initialized by the rated parameters on the nameplate of the permanent magnet synchronous motor or the electrical parameters measured by offline tests.

[0023] Preferably, in step 1), the expression of the adjustable mathematical model of the permanent magnet synchronous motor in the dq rotating coordinate system is as follows:

[0024] ;

[0025] In the formula, This is the estimated value of the d-axis stator current. This is an estimated value for the stator synchronous inductance. The d-axis stator voltage component. This is an estimated value for the stator resistance. The d-axis stator current component. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the estimated value of the q-axis stator current. The q-axis stator voltage component This is the estimated value of the flux linkage of the rotor permanent magnet.

[0026] Preferably, in step 4), the current error signal includes d-axis current error and q-axis current error, wherein the d-axis current error is the difference between the real-time acquired d-axis stator current component and the estimated d-axis stator current value, and the q-axis current error is the difference between the real-time acquired q-axis stator current component and the estimated q-axis stator current value.

[0027] Preferably, in step 5), the parallel adaptive update law adopts a proportional-integral adjustment structure, wherein the stator resistance estimate, stator synchronous inductance estimate, and rotor permanent magnet flux linkage estimate each correspond to an independent update formula:

[0028] Stator resistance estimate update formula:

[0029] ;

[0030] In the formula, For the stator resistance estimate that is updated in real time, Identify the proportional gain coefficient for the stator resistance. Identify the integral gain coefficient for the stator resistance. For the Laplace complex frequency domain operator, This is the d-axis current error signal. This is the q-axis current error signal. The d-axis stator current component. This refers to the q-axis stator current component;

[0031] Stator synchronous inductance estimate update formula:

[0032] ;

[0033] In the formula, This is an estimate of the stator synchronous inductance that is updated in real time. To identify the proportional gain coefficient of the stator synchronous inductor, Identify the integral gain coefficient for the stator synchronous inductor. For the Laplace complex frequency domain operator, This is the d-axis current error signal. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the q-axis current error signal. The d-axis stator current component;

[0034] Update formula for rotor permanent magnet flux linkage estimate:

[0035] ;

[0036] In the formula, This is an estimate of the rotor permanent magnet flux linkage that is updated in real time. To identify the proportional gain coefficient for rotor flux linkage, Identify the integral gain coefficient for the rotor flux linkage. For the Laplace complex frequency domain operator, This is the q-axis current error signal. ω is the rotor's electric angular velocity.

[0037] Preferably, in step 5), the parameter update cycle of the parallel adaptive update law is completely consistent with the current loop control cycle of the permanent magnet synchronous motor field orientation control system, so as to realize the synchronous iterative execution of parameter identification and motor control.

[0038] Preferably, in step 6), the method for determining the convergence of the current error signal is as follows: set the d-axis error threshold and the q-axis error threshold. When the absolute value of the d-axis current error is continuously less than the d-axis error threshold and the absolute value of the q-axis current error is continuously less than the q-axis error threshold in multiple consecutive control cycles, the current error signal is determined to converge, and the identification parameters obtained in the current iteration are output as the effective electrical parameters of the motor.

[0039] The beneficial effects of this invention are as follows:

[0040] A model-reference adaptive online parameter identification method for surface-mounted permanent magnet synchronous motors, executed cyclically according to the control cycle of the field-oriented control system, includes the following steps:

[0041] 1) Based on the premise that the d-axis inductance and the q-axis inductance are equal, construct an adjustable mathematical model of the permanent magnet synchronous motor in the dq rotating coordinate system containing the set of parameters to be identified. The set of parameters to be identified includes the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage.

[0042] 2) During each control cycle of normal motor operation, the stator voltage component, stator current component, and rotor electric angular velocity in the dq rotating coordinate system are collected in real time. The normal operation process of the motor is the d-axis stator current. Furthermore, the steady-state operation of the field-oriented control system is characterized by no drastic changes in motor speed and load torque.

[0043] 3) Using the stator voltage component, stator current component, rotor electric angular velocity, and the parameter set to be identified updated in the previous control cycle obtained in step 2), input the adjustable mathematical model of the permanent magnet synchronous motor to calculate the estimated value of the stator current.

[0044] 4) The difference between the real-time acquired stator current components and the estimated stator current is calculated to generate a current error signal;

[0045] 5) Construct a parallel adaptive update law. The parallel adaptive update law takes the stator current component and rotor electric angular velocity obtained in real time in step 2) and the current error signal obtained in step 4) as inputs. Under the premise of not changing the current operating conditions of the motor, it synchronously updates the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage.

[0046] 6) Determine whether the current error signal has converged. If it has converged, then output the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage obtained by parallel adaptive update law iteration as the actual electrical parameters of the motor under the current operating condition.

[0047] In existing technologies, offline identification schemes require the motor to be shut down and a dedicated excitation signal to be injected, making it impossible to track electrical parameter drift caused by winding temperature rise and magnetic saturation effects during motor operation. Traditional single-parameter online identification methods, on the other hand, struggle to fully characterize the actual operating state of the motor. In contrast, this invention constructs an adjustable mathematical model in a dq rotating coordinate system under normal motor operating conditions, incorporating estimated stator resistance, estimated stator synchronous inductance, and estimated rotor permanent magnet flux linkage. The model inputs are the stator voltage component, stator current component, and rotor electric angular velocity already collected in the actual control system. The error signal between the estimated stator current output from the model and the measured current drives a parallel adaptive update law to synchronously iterate the three parameters to be identified. In other words, this invention allows for real-time tracking of the dynamic changes of three key electrical parameters—stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage—under motor load conditions without requiring motor shutdown or additional high-frequency excitation signals to the control system. The update channels corresponding to each parameter are structurally decoupled, effectively suppressing coupling interference during multi-parameter synchronous identification. Meanwhile, the entire identification process does not interfere with or change the current operating conditions of the motor, thus fundamentally avoiding the torque pulsation and electromagnetic noise problems caused by external excitation signals. Furthermore, this invention uses the convergence of the current error signal as the output criterion for valid identification results, ensuring the reliability of the output electrical parameters.

[0048] Preferably, in step 1), while constructing the adjustable mathematical model of the permanent magnet synchronous motor, the set of parameters to be identified is initialized. The initial values ​​of the set of parameters to be identified are assigned and initialized by the rated parameters on the nameplate of the permanent magnet synchronous motor or the electrical parameters measured by offline tests.

[0049] This invention provides a reasonable initial starting point, close to the actual motor parameters, for the iterative process of the parallel adaptive update law by initializing the parameter set to be identified using the rated parameters on the motor nameplate or electrical parameters measured offline during the construction of an adjustable mathematical model. Compared to setting the initial parameter values ​​to zero or arbitrarily randomized values, this initialization method effectively shortens the convergence path of the adaptive iteration, enabling the identification results to reach the preset error convergence threshold within a shorter number of control cycles. Simultaneously, it effectively avoids the problem of drastic fluctuations or even divergence of the current error signal during the iteration process due to excessive deviation between the initial values ​​and the actual parameters, ensuring the numerical stability of the parameter identification process in the startup phase and laying a reliable foundation for the smooth iteration of the adaptive update law in subsequent control cycles.

[0050] Preferably, the normal operating process of the motor is the d-axis stator current. Furthermore, the steady-state operation of the field-oriented control system is characterized by no drastic changes in motor speed and load torque.

[0051] This invention utilizes the d-axis stator current Furthermore, parameter identification is performed under steady-state operation conditions of field-oriented control where motor speed and load torque do not change drastically. Under these conditions, the d-axis stator current component remains zero, and the entire load torque of the motor is borne by the q-axis stator current component. Under these conditions, the cross-coupling terms between the d-axis and q-axis in the adjustable mathematical model are significantly simplified. The d-axis current error signal mainly reflects the parameter deviation information of the stator resistance, while the q-axis current error signal mainly reflects the parameter deviation information of the rotor permanent magnet flux linkage. This provides physically meaningful and approximately orthogonal identification excitations for each parameter channel in the parallel adaptive update law. This characteristic helps improve the decoupling level of each parameter update channel and enhances the identification accuracy of multi-parameter synchronous identification. Simultaneously, the steady-state operation condition ensures the stability of voltage, current, and speed feedback signals, avoiding interference from transient disturbances on the adaptive iteration process, and providing favorable operating conditions for the convergence stability of parameter identification.

[0052] Preferably, in step 1), the expression of the adjustable mathematical model of the permanent magnet synchronous motor in the dq rotating coordinate system is as follows:

[0053] ;

[0054] In the formula, This is the estimated value of the d-axis stator current. This is an estimated value for the stator synchronous inductance. The d-axis stator voltage component. This is an estimated value for the stator resistance. The d-axis stator current component. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the estimated value of the q-axis stator current. The q-axis stator voltage component This is the estimated value of the flux linkage of the rotor permanent magnet.

[0055] This invention constructs an adjustable mathematical model of a permanent magnet synchronous motor in the dq rotating coordinate system. This model can accurately reflect the performance of a surface-mounted permanent magnet synchronous motor. Electromagnetic dynamic characteristics under certain conditions. In this model, the estimated stator resistance value... The resistive voltage drop term that primarily affects the current, and the estimated value of the stator synchronous inductance. Simultaneously acting on the inertial element and rotational coupling term of the voltage equation, the estimated value of the rotor permanent magnet flux linkage... The back electromotive force term acting on the q-axis is used. The three parameters to be identified are distributed differently within the model structure, allowing the subsequent parallel adaptive update law to design driving terms with clear physical meanings for each parameter, providing a mathematical basis for decoupling and identifying the parameters. Simultaneously, the model outputs the stator current estimate, which can be directly compared with the measured stator current to generate a current error signal, providing a driving force with a clear convergence direction for the adaptive update law. The model contains only first-order differential equations, resulting in low computational overhead, and can complete iterative solutions within each control cycle of a general-purpose digital computing chip, ensuring the engineering feasibility of the entire online identification process.

[0056] Preferably, in step 4), the current error signal includes d-axis current error and q-axis current error, wherein the d-axis current error is the difference between the real-time acquired d-axis stator current component and the estimated d-axis stator current value, and the q-axis current error is the difference between the real-time acquired q-axis stator current component and the estimated q-axis stator current value.

[0057] This invention defines d-axis current error and q-axis current error separately, and independently represents the deviation between the estimated stator current value and the measured stator current component in two orthogonal axes of the dq rotating coordinate system. Compared with the processing method that only uses a single synthesized current error amplitude, this invention's dual-axis error signal completely preserves the directional information of the error in two orthogonal axes, and can provide physically meaningful and mutually orthogonal driving quantities for the three parameter channels of stator resistance, stator synchronous inductance and rotor permanent magnet flux linkage in the parallel adaptive update law.

[0058] Among them, the d-axis current error mainly carries the deviation information of the stator resistance, and the q-axis current error mainly carries the deviation information of the rotor permanent magnet flux linkage. The error signals in the two axes are combined with the current component and the speed, and then input into the update formula corresponding to each parameter. This gives the three parameters to be identified different identification excitation sources, further improving the decoupling level between the update channels of each parameter, and laying the necessary signal foundation for achieving the convergence accuracy and system stability of multi-parameter synchronous identification.

[0059] Preferably, in step 5), the parallel adaptive update law adopts a proportional-integral adjustment structure, wherein the stator resistance estimate, stator synchronous inductance estimate, and rotor permanent magnet flux linkage estimate each correspond to an independent update formula:

[0060] Stator resistance estimate update formula:

[0061] ;

[0062] In the formula, For the stator resistance estimate that is updated in real time, Identify the proportional gain coefficient for the stator resistance. Identify the integral gain coefficient for the stator resistance. For the Laplace complex frequency domain operator, This is the d-axis current error signal. This is the q-axis current error signal. The d-axis stator current component. This refers to the q-axis stator current component;

[0063] Stator synchronous inductance estimate update formula:

[0064] ;

[0065] In the formula, This is an estimate of the stator synchronous inductance that is updated in real time. To identify the proportional gain coefficient of the stator synchronous inductor, Identify the integral gain coefficient for the stator synchronous inductor. For the Laplace complex frequency domain operator, This is the d-axis current error signal. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the q-axis current error signal. The d-axis stator current component;

[0066] Update formula for rotor permanent magnet flux linkage estimate:

[0067] ;

[0068] In the formula, This is an estimate of the rotor permanent magnet flux linkage that is updated in real time. To identify the proportional gain coefficient for rotor flux linkage, Identify the integral gain coefficient for the rotor flux linkage. For the Laplace complex frequency domain operator, This is the q-axis current error signal. ω is the rotor's electric angular velocity.

[0069] This invention employs a proportional-integral (PI) control structure, designing independent update formulas for stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage. The proportional term enables rapid response to current errors, allowing the parameter estimates to quickly approach the true values ​​in the initial identification phase. The integral term eliminates steady-state errors, ensuring steady-state identification accuracy. These two terms work synergistically to balance dynamic response speed and steady-state identification accuracy.

[0070] In other words, the stator resistance update formula uses the sum of the products of the d-axis and q-axis current errors and their corresponding current components as the driving term, characterizing the impact of resistance deviation on copper loss voltage drop. The stator synchronous inductance update formula uses a cross-coupled driving term composed of error, speed, and current, reflecting the effect of inductance deviation on the induced electromotive force of the rotating coordinate system. The rotor permanent magnet flux linkage update formula uses the product of the q-axis current error and speed as the driving term, reflecting the direct impact of flux linkage deviation on the q-axis back electromotive force. Each driving term corresponds to a different physical action term in the adjustable mathematical model, enabling structural decoupling of the parameter update channels and avoiding coupling interference and convergence oscillations during multi-parameter synchronous identification. Furthermore, all update formulas reuse only existing acquired signals, requiring no additional excitation or special operating conditions, providing core algorithmic support for achieving high-precision synchronous online identification of three key electrical parameters without disturbing the original normal operating state of the motor.

[0071] Preferably, in step 5), the parameter update cycle of the parallel adaptive update law is completely consistent with the current loop control cycle of the permanent magnet synchronous motor field orientation control system, so as to realize the synchronous iterative execution of parameter identification and motor control.

[0072] This invention sets the parameter update cycle of the parallel adaptive update law to be completely consistent with the current loop control cycle of the field-oriented control system, enabling the parameter identification iteration process and the motor current regulation process to run synchronously within the same timing framework. Compared to implementations using independent timers to trigger identification or using slow identification cycles lower than the current loop operating frequency, this invention ensures that the parameters to be identified used in the adjustable mathematical model calculations within each control cycle are the latest estimated values ​​obtained from the iteration. In other words, the parameter estimates obtained by this invention can track the changes in electrical parameters of the motor caused by operating conditions such as temperature rise and magnetic saturation in real time at the same refresh rate as the current loop, avoiding the risks of model mismatch, current error amplification, and even system instability caused by the identification update lagging behind the control update.

[0073] Preferably, in step 6), the method for determining the convergence of the current error signal is as follows: set the d-axis error threshold and the q-axis error threshold. When the absolute value of the d-axis current error is continuously less than the d-axis error threshold and the absolute value of the q-axis current error is continuously less than the q-axis error threshold in multiple consecutive control cycles, the current error signal is determined to converge, and the identification parameters obtained in the current iteration are output as the effective electrical parameters of the motor.

[0074] This invention sets independent error thresholds on two orthogonal axes of the dq rotating coordinate system and adds a time-duration judgment condition of "multiple consecutive control cycles". On the one hand, compared with the method of directly using a single synthetic current error amplitude for convergence judgment, the dual-axis independent threshold judgment can monitor the convergence status of the current errors of the d-axis and q-axis respectively. This avoids the defect of prematurely ending the identification when the synthetic error amplitude meets the condition, even though the error of one axis has been suppressed, while the error of the other axis is still too large. This ensures that the identification accuracy of the two coordinate axes of d and q simultaneously meets the index requirements, improving the completeness of the convergence criterion. On the other hand, the time-duration constraint of "multiple consecutive control cycles" can effectively filter out the phenomenon of accidental instantaneous drops in current error caused by load disturbances, sampling noise, or transient operating condition fluctuations. This avoids the adaptive update law outputting the identification result prematurely due to the accidental decrease of a single error before the parameters have truly converged to the steady-state value. This decision mechanism ensures that the identified parameters are only output when the parameter estimates have been sufficiently iterated and remain stable near the actual parameters. This guarantees the reliability and identification accuracy of the final output stator resistance estimate, stator synchronous inductance estimate, and rotor permanent magnet flux linkage estimate, providing a reliable parameter input basis for the subsequent controller to carry out adaptive correction based on the identification results.

[0075] Definitions:

[0076] Parallel adaptive update law: In this invention, it refers to a proportional-integral adaptive update algorithm that synchronously updates three parameters to be identified—stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage—based on the same set of current error signals without switching motor operating conditions. Attached Figure Description

[0077] Figure 1 This is a flowchart of the present invention;

[0078] Figure 2 This is a system block diagram of the present invention;

[0079] Figure 3 This is a structural block diagram of the MRAS parameter identification module in this invention;

[0080] Figure 4 This is a schematic diagram of the simulation results curve for multi-parameter synchronous identification in an embodiment of the present invention. Detailed Implementation

[0081] Example 1: Online Parameter Identification System for Surface-Mounted Permanent Magnet Synchronous Motors Based on Model Reference Adaptation

[0082] like Figures 2 to 3As shown, this embodiment provides an online parameter identification system for surface-mount permanent magnet synchronous motors. The system is integrated into a digital computing chip to realize online identification of key electrical parameters of surface-mount permanent magnet synchronous motors (SPMSM) during operation, such as digital signal processors (DSPs), field programmable gate arrays (FPGAs), and system-on-chips (SoCs).

[0083] In this invention, the system mainly includes a signal acquisition unit, a coordinate transformation unit, a field orientation control unit, and an online parameter identification unit. All units work collaboratively and execute synchronously in a cyclical manner according to the current loop control cycle of the field orientation control (FOC) system, wherein:

[0084] (1) Signal acquisition unit: used to acquire raw physical signals during motor operation in real time, including current acquisition subunit, position and speed acquisition subunit and voltage acquisition subunit.

[0085] Current acquisition subunit: Located between the inverter and the motor stator windings, used to acquire three-phase stator current. ;

[0086] Position and speed acquisition subunit: Connected to the position sensor (such as encoder or rotary transformer) on the motor shaft, used to acquire the rotor's mechanical position angle. The mechanical angular velocity was obtained by calculating the time derivative. Furthermore, this is combined with the number of motor pole pairs. The rotor electric angular velocity was calculated. ;

[0087] Voltage acquisition subunit: used to acquire three-phase stator voltage. In practical applications, voltage data can be directly acquired via a voltage sensor or obtained using voltage commands output from a magnetic field orientation control unit. It is obtained by inverse transformation and reconstruction.

[0088] The three-phase voltage signal is output to the coordinate transformation unit through three methods: current loop voltage command reconstruction, or a combination of inverter model and dead-zone compensation reconstruction.

[0089] (2) Coordinate Transformation Unit

[0090] The coordinate transformation unit receives the electrical signal in the three-phase stationary coordinate system output by the signal acquisition unit and transforms it to a two-phase rotating coordinate system (dq axis). This coordinate transformation unit includes a Clark transformation subunit and a Park transformation subunit, wherein:

[0091] Clark transform subunit: converts three-phase stator current Three-phase voltage Transform to a two-phase stationary coordinate system ( (Axis) below.

[0092] Park transformation subunit: combined with rotor electrical angle (Depend on (Points earned), will soon The current and voltage signals in the d-axis coordinate system are further transformed into stator current components in the dq rotating coordinate system. With stator voltage components The converted dq-axis electrical signals are synchronously output to the magnetic field orientation control unit and the online parameter identification unit.

[0093] (3) Magnetic field orientation control unit

[0094] The field-oriented control unit constitutes the main closed-loop control circuit of the motor, ensuring stable operation of the motor under set operating conditions. In this embodiment, the motor adopts... The field-oriented control strategy comprises a velocity loop PI subunit, a current loop PI subunit, a Park inverse transform subunit, and an SVPWM subunit.

[0095] Speed ​​loop PI subunit: Receives rotational speed command With feedback rotor electrical angular velocity The output q-axis current setpoint is adjusted by proportional-integral (PI) regulation. .

[0096] Current loop PI subunit: Receives d-axis current setpoint (Constantly set to 0) and q-axis current setpoint and feedback The comparison is performed, and the output dq axis voltage command is then adjusted by a PI controller. This voltage command can be directly used as a parameter to identify the stator voltage component required by the unit online. .

[0097] Park inverse converter subunit and SVPWM subunit: Restore the dq voltage command to... The coordinate system voltage command (three-phase voltage signal) is used to generate six pulse width modulation (PWM) signals to drive the inverter and control the motor operation.

[0098] (4) Online parameter identification unit

[0099] The online parameter identification unit is used to identify the stator resistance in real time without interfering with the normal operation of the motor. Stator synchronous inductor and rotor permanent magnet flux The online parameter identification unit includes an adjustable mathematical model subunit, an error calculation subunit, an adaptive law subunit, and a parameter output and application subunit.

[0100] Adjustable mathematical model sub-unit: Used to construct an adjustable mathematical model of the permanent magnet synchronous motor in the dq rotating coordinate system, containing the set of parameters to be identified. Its input is the output of the coordinate transformation unit. , And the set of parameters to be identified that has been updated in the previous control cycle ( , , ).

[0101] The adjustable mathematical model subunit uses the forward Euler method as the numerical discrete integration algorithm, and performs iterative calculations based on the built-in motor discrete state equations to solve and output the estimated values ​​of the d-axis and q-axis stator currents corresponding to the current control cycle. , The discrete iterative mathematical expressions corresponding to the d-axis and q-axis are as follows:

[0102] for axis:

[0103]

[0104] In the formula, For the first Estimated d-axis stator current during control period This is the estimated value of the d-axis stator current during the k-th control cycle. The control cycle of the magnetic field orientation control system. This is an estimated value for the stator synchronous inductance. For the d-axis stator voltage component of the k-th control period, This is an estimated value for the stator resistance. For the d-axis stator current component in the k-th control period, For the rotor electric angular velocity in the kth control cycle, This represents the q-axis stator current component during the k-th control cycle.

[0105] for axis:

[0106]

[0107] In the formula, For the first Estimated q-axis stator current during the control period. This is the estimated value of the q-axis stator current during the k-th control cycle. The control cycle of the magnetic field orientation control system. This is an estimated value for the stator synchronous inductance. For the q-axis stator voltage component of the k-th control cycle, This is an estimated value for the stator resistance. For the q-axis stator current component in the k-th control period, For the rotor electric angular velocity in the kth control cycle, For the d-axis stator current component in the k-th control period, This is the estimated value of the flux linkage of the rotor permanent magnet.

[0108] The mathematical model expression built into the adjustable mathematical model subunit is as follows:

[0109]

[0110] In the formula, This is the estimated value of the d-axis stator current. This is an estimated value for the stator synchronous inductance. The d-axis stator voltage component. This is an estimated value for the stator resistance. The d-axis stator current component. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the estimated value of the q-axis stator current. The q-axis stator voltage component This is the estimated value of the flux linkage of the rotor permanent magnet.

[0111] The adjustable mathematical model sub-unit calculates the estimated stator current for the current cycle using numerical integration methods (such as the forward Euler method). , ).

[0112] Error calculation subunit: Receives the actual stator current component output by the coordinate transformation unit ( The stator current estimate output by the adjustable mathematical model sub-unit () , The difference operation is then performed to generate a current error signal. Specifically, the d-axis current error is: The q-axis current error is: .

[0113] Adaptive Law Subunit: This subunit employs a parallel adaptive update law based on the current error signal. , Real-time acquisition of stator current components and rotor electric angular velocity As input, three parameters to be identified are updated synchronously. This adaptive law subunit internally sets up three independent PI controllers, with the update formula as follows:

[0114] Stator resistance estimate update formula:

[0115] ;

[0116] In the formula, For the stator resistance estimate that is updated in real time, Identify the proportional gain coefficient for the stator resistance. Identify the integral gain coefficient for the stator resistance. For the Laplace complex frequency domain operator, This is the d-axis current error signal. This is the q-axis current error signal. The d-axis stator current component. This refers to the q-axis stator current component;

[0117] Stator synchronous inductance estimate update formula:

[0118] ;

[0119] In the formula, This is an estimate of the stator synchronous inductance that is updated in real time. To identify the proportional gain coefficient of the stator synchronous inductor, Identify the integral gain coefficient for the stator synchronous inductor. For the Laplace complex frequency domain operator, This is the d-axis current error signal. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the q-axis current error signal. The d-axis stator current component;

[0120] Update formula for rotor permanent magnet flux linkage estimate:

[0121] ;

[0122] In the formula, This is an estimate of the rotor permanent magnet flux linkage that is updated in real time. To identify the proportional gain coefficient for rotor flux linkage, Identify the integral gain coefficient for the rotor flux linkage. For the Laplace complex frequency domain operator, This is the q-axis current error signal. ω is the rotor's electric angular velocity.

[0123] Parameter Output and Application Subunit: This subunit is responsible for determining whether the identification process has converged and outputting the final parameters. The built-in convergence determination logic is as follows: Preset d-axis and q-axis error thresholds are used. The absolute value of the d-axis current error is determined over multiple consecutive control cycles. The absolute value of the q-axis current error is consistently less than the d-axis error threshold. When the current error signal remains consistently below the q-axis error threshold, it is considered converged. At this point, the current error signal obtained in the current iteration is... , , This output serves as the actual electrical parameter output for the motor under its current operating conditions. The output parameters can be fed back to the field-oriented control unit for online correction of controller parameters, thereby improving control performance.

[0124] Example 2: Online Parameter Identification Method for Surface-Mounted Permanent Magnet Synchronous Motors Based on Model Reference Adaptation

[0125] This embodiment provides an online parameter identification method for surface-mounted permanent magnet synchronous motors based on model reference adaptation. This method is executed using the system hardware platform described in Embodiment 1. Its core lies in constructing a model reference adaptive system (MRAS) that runs parallel to the actual control loop, enabling real-time identification of the motor's stator resistance, stator inductance, and rotor flux linkage without injecting additional high-frequency signals.

[0126] like Figures 1 to 3 As shown, this method follows the current loop control cycle of the field-oriented control (FOC) system. The process is repeated continuously. The specific steps are as follows:

[0127] 1) Based on the premise that the d-axis inductance and the q-axis inductance are equal, construct an adjustable mathematical model of the permanent magnet synchronous motor in the dq rotating coordinate system containing the set of parameters to be identified.

[0128] During the system power-on or reset phase, an adjustable mathematical model of the permanent magnet synchronous motor is first constructed, and the set of parameters to be identified is initialized.

[0129] Specifically, the premise for constructing this adjustable mathematical model of the permanent magnet synchronous motor is: based on the physical characteristics of the surface-mounted permanent magnet synchronous motor (SPMSM), the d-axis inductance is set to be equal to the q-axis inductance, that is... .

[0130] Based on the rated parameters on the motor nameplate or the electrical parameters measured through offline testing, parameter initialization is performed, that is, setting the initial values ​​of the set of parameters to be identified. , , For example, if the nameplate indicates a stator resistance of... Then initialize .

[0131] In the dq rotating coordinate system, the following state equations are established as the adjustable mathematical model of the permanent magnet synchronous motor:

[0132]

[0133] In the formula, This is the estimated value of the d-axis stator current. This is an estimated value for the stator synchronous inductance. The d-axis stator voltage component. This is an estimated value for the stator resistance. The d-axis stator current component. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the estimated value of the q-axis stator current. The q-axis stator voltage component This is the estimated value of the flux linkage of the rotor permanent magnet.

[0134] 2) During each control cycle of normal motor operation (in this embodiment, the motor is maintained) (For the steady-state operation of magnetic field orientation control), the following signals are acquired in real time through the signal acquisition unit and coordinate transformation unit:

[0135] Stator voltage components in the dq rotating coordinate system , (Usually taken from the output command of the current loop PI regulator);

[0136] Stator current components in the dq rotating coordinate system (Acquired by a current sensor and converted using Clark / Park conversion);

[0137] Rotor electric angular velocity (Obtained by differentiating the angle measured by the position sensor).

[0138] 3) Utilize the stator voltage component obtained in step 2) , With rotor electric angular velocity and the set of parameters to be identified that was updated in the previous control cycle. Input into the adjustable mathematical model of the permanent magnet synchronous motor.

[0139] To avoid noise interference from direct differentiation, this embodiment uses a discretization iterative algorithm (such as the forward Euler method) to solve the above differential equation and calculate the estimated value of the stator current for the current period.

[0140] 4) The stator current components acquired in real time in step 2) The current error signal is generated by subtracting the estimated stator current obtained in step 3) from the calculated value. The specific calculation formula is as follows:

[0141]

[0142] In the formula, For d-axis current error, The d-axis stator current component. This is the estimated value of the d-axis stator current. For q-axis current error, This represents the q-axis stator current component. This is the estimated value of the q-axis stator current.

[0143] The aforementioned current error signal can effectively reflect the dynamic difference between the adjustable mathematical model and the actual motor (reference model).

[0144] If the motor enters Under weak field operation conditions, the motor's magnetic circuit will be in a state of deep saturation. On the one hand, magnetic saturation will cause a deviation in the actual inductance parameters of the motor, resulting in current error signals. On the one hand, the accuracy decreases; on the other hand, magnetic saturation exacerbates iron loss, directly leading to instability of the identification equation set. Therefore, the identification equation set in this embodiment is only suitable for parameter identification under the rated operating conditions of the motor, and needs to be limited to... The magnetic field orientation steady-state operation condition is used for identification, thereby avoiding the risk of parameter deviation and algorithm instability caused by magnetic saturation, and ensuring identification accuracy and operational stability.

[0145] 5) Construct a parallel adaptive update law, wherein the parallel adaptive update law is based on the stator current components acquired in real time in step 2). and rotor electric angular velocity The current error signal obtained in step 4) Using a proportional-integral (PI) control structure as input, the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage are updated synchronously without changing the current operating conditions of the motor. The specific update formula is as follows:

[0146] Stator resistance estimate update formula:

[0147] ;

[0148] In the formula, For the stator resistance estimate that is updated in real time, Identify the proportional gain coefficient for the stator resistance. Identify the integral gain coefficient for the stator resistance. For the Laplace complex frequency domain operator, This is the d-axis current error signal. This is the q-axis current error signal. The d-axis stator current component. This refers to the q-axis stator current component;

[0149] By using the product of current error and current component to drive the correction of resistance parameters, the stator resistance estimate can be monotonically converged in the direction of decreasing error energy. While ensuring the global stability of parameter identification, there is no need to inject additional high-frequency excitation signals into the motor, thus avoiding interference with the normal operation of the motor during the identification process.

[0150] Stator synchronous inductance estimate update formula:

[0151] ;

[0152] In the formula, This is an estimate of the stator synchronous inductance that is updated in real time. To identify the proportional gain coefficient of the stator synchronous inductor, Identify the integral gain coefficient for the stator synchronous inductor. For the Laplace complex frequency domain operator, This is the d-axis current error signal. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the q-axis current error signal. This represents the d-axis stator current component.

[0153] By using the product of error and the cross-coupling terms of speed and current to drive the correction of inductance parameters, the cross-coupling characteristics of stator inductance in two axes under the dq rotating coordinate system can be fully utilized. This ensures that the estimated value of stator synchronous inductance continuously receives sufficient parameter identification excitation during motor rotation, thereby guaranteeing the convergence and accuracy of inductance identification. At the same time, it is independent of the update channels of stator resistance and rotor flux linkage, avoiding coupling interference between multiple parameters.

[0154] Update formula for rotor permanent magnet flux linkage estimate:

[0155] ;

[0156] In the formula, This is an estimate of the rotor permanent magnet flux linkage that is updated in real time. To identify the proportional gain coefficient for rotor flux linkage, Identify the integral gain coefficient for the rotor flux linkage. For the Laplace complex frequency domain operator, This is the q-axis current error signal. Let be the rotor electrical angular velocity. This formula mainly uses the product of the q-axis current error and the rotational speed to drive the correction of the flux linkage parameters. It can accurately reflect the influence mechanism of the rotor permanent magnet flux linkage on the q-axis back electromotive force. Under the condition that the motor is rotating and the speed is not zero, it provides sufficient excitation conditions for flux linkage identification, so that the estimated value of the rotor permanent magnet flux linkage converges in the direction of decreasing error energy. At the same time, this update channel only depends on the q-axis error signal and is decoupled from the update channels of stator resistance and stator synchronous inductance, avoiding mutual interference in the multi-parameter synchronous identification process.

[0157] The three update formulas described above are executed in parallel, and their update cycles are completely consistent with the current loop control cycle of the FOC system. The parallel adaptive update law described in this invention is essentially a proportional-integral (PI) regulator structure, and the above six sets of gain coefficients... , , , , , The gain coefficient directly determines whether the identification algorithm can converge, i.e., the identification stability, convergence speed, and the decoupling effect between the stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage. The gain coefficient needs to be selected comprehensively based on the motor's rated power level and the control system's control cycle; in a specific embodiment, when the control cycle... In other words, under the control frequency of 100kHz, the gain values ​​for each identification channel are: stator resistance identification channel proportional gain. Stator resistance identification integral gain coefficient Stator synchronous inductor identification proportional gain coefficient Stator synchronous inductor identification integral gain coefficient Rotor flux identification proportional gain coefficient Rotor flux identification integral gain coefficient .

[0158] The gain coefficients of this set can be tuned through simulation analysis or bench tests. The tuning principle is: under the premise of ensuring that the three identification channels of stator resistance, stator synchronous inductance and rotor permanent magnet flux are all stably converged, the integral gain should be increased first to improve the steady-state accuracy of parameter identification. At the same time, the proportional gain should be adjusted to suppress the parameter oscillation phenomenon generated during the identification convergence process.

[0159] 6) The system continuously monitors the current error signal. The state of the current error signal is used to determine whether it has converged.

[0160] Preset d-axis error threshold and q-axis error threshold (For example, all are set to 0.1A). When continuous One control cycle (e.g.) Within ) to satisfy and When the current error signal converges, it is determined that the signal has converged.

[0161] Once converged, the system will obtain the results from the current iteration. The actual electrical parameters of the motor under the current operating conditions are locked and output. These parameters can be directly fed back to the FOC controller to correct the current loop PI parameters or perform field weakening control, thereby achieving self-adaptation of the control system.

[0162] 7) After completing the parameter update and output determination for the current cycle, the process automatically returns to step 2) and enters the next control cycle. If the motor operating state changes (such as a change in resistance due to increased temperature), the error signal will increase again, and the adaptive law module will automatically readjust the parameters until convergence occurs again, thereby achieving online identification under all operating conditions.

[0163] The controller and control method described above were used to build a simulation model for verification. The motor model parameters were designed as follows: stator synchronous inductance... stator resistance Rotor permanent magnet flux During the identification process, all three parameters to be identified start from initial values ​​that deviate from the true values. After iteration using a parallel adaptive update law, they are... It then entered steady-state convergence. The final identification result was: stator synchronous inductor. stator resistance Rotor permanent magnet flux The identification accuracy meets the requirements of high-performance control, such as Figure 4 As shown.

[0164] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications made to the present invention by those skilled in the art without departing from the spirit of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for online parameter identification of a surface-mounted permanent magnet synchronous motor based on model reference adaptation, which is executed cyclically according to the control cycle of a field-oriented control system, characterized in that... Includes the following steps: 1) Based on the premise that the d-axis inductance and the q-axis inductance are equal, construct an adjustable mathematical model of the permanent magnet synchronous motor in the dq rotating coordinate system containing the set of parameters to be identified. The set of parameters to be identified includes the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage. 2) During each control cycle of normal motor operation, the stator voltage component, stator current component, and rotor electric angular velocity in the dq rotating coordinate system are collected in real time. The normal operation process of the motor is the d-axis stator current. Furthermore, the steady-state operation of the field-oriented control system is characterized by no drastic changes in motor speed and load torque. 3) Using the stator voltage component, stator current component, rotor electric angular velocity, and the parameter set to be identified updated in the previous control cycle obtained in step 2), input the adjustable mathematical model of the permanent magnet synchronous motor to calculate the estimated value of the stator current. 4) The difference between the real-time acquired stator current components and the estimated stator current is calculated to generate a current error signal; 5) Construct a parallel adaptive update law. The parallel adaptive update law takes the stator current component and rotor electric angular velocity obtained in real time in step 2) and the current error signal obtained in step 4) as inputs. Under the premise of not changing the current operating conditions of the motor, it synchronously updates the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage. 6) Determine whether the current error signal has converged. If it has converged, then output the estimated values ​​of stator resistance, stator synchronous inductance, and rotor permanent magnet flux linkage obtained by parallel adaptive update law iteration as the actual electrical parameters of the motor under the current operating condition.

2. The method for online parameter identification of a surface-mounted permanent magnet synchronous motor according to claim 1, characterized in that, In step 1), while constructing the adjustable mathematical model of the permanent magnet synchronous motor, the set of parameters to be identified is initialized. The initial values ​​of the set of parameters to be identified are assigned and initialized by the rated parameters on the nameplate of the permanent magnet synchronous motor or the electrical parameters measured by offline tests.

3. The method for online parameter identification of a surface-mounted permanent magnet synchronous motor according to claim 1, characterized in that, In step 1), the expression of the adjustable mathematical model of the permanent magnet synchronous motor in the dq rotating coordinate system is as follows: ； In the formula, This is the estimated value of the d-axis stator current. This is an estimated value for the stator synchronous inductance. The d-axis stator voltage component. This is an estimated value for the stator resistance. The d-axis stator current component. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the estimated value of the q-axis stator current. The q-axis stator voltage component This is the estimated value of the rotor permanent magnet flux linkage.

4. The method for online parameter identification of a surface-mounted permanent magnet synchronous motor according to claim 1, characterized in that, In step 4), the current error signal includes d-axis current error and q-axis current error, wherein the d-axis current error is the difference between the real-time acquired d-axis stator current component and the estimated d-axis stator current value, and the q-axis current error is the difference between the real-time acquired q-axis stator current component and the estimated q-axis stator current value.

5. The method for online parameter identification of a surface-mounted permanent magnet synchronous motor according to claim 1, characterized in that, In step 5), the parallel adaptive update law adopts a proportional-integral adjustment structure, wherein the stator resistance estimate, stator synchronous inductance estimate, and rotor permanent magnet flux linkage estimate each correspond to an independent update formula: Stator resistance estimate update formula: ； In the formula, For the stator resistance estimate that is updated in real time, Identify the proportional gain coefficient for the stator resistance. Identify the integral gain coefficient for the stator resistance. For the Laplace complex frequency domain operator, This is the d-axis current error signal. This is the q-axis current error signal. The d-axis stator current component. This refers to the q-axis stator current component; Stator synchronous inductance estimate update formula: ； In the formula, This is an estimate of the stator synchronous inductance that is updated in real time. To identify the proportional gain coefficient of the stator synchronous inductor, Identify the integral gain coefficient for the stator synchronous inductor. For the Laplace complex frequency domain operator, This is the d-axis current error signal. The rotor's electric angular velocity, This represents the q-axis stator current component. This is the q-axis current error signal. The d-axis stator current component; Update formula for rotor permanent magnet flux linkage estimate: ； In the formula, This is an estimate of the rotor permanent magnet flux linkage that is updated in real time. To identify the proportional gain coefficient for rotor flux linkage, Identify the integral gain coefficient for the rotor flux linkage. For the Laplace complex frequency domain operator, This is the q-axis current error signal. ω is the rotor's electric angular velocity.

6. The method for online parameter identification of a surface-mounted permanent magnet synchronous motor according to claim 1, characterized in that, In step 5), the parameter update cycle of the parallel adaptive update law is completely consistent with the current loop control cycle of the permanent magnet synchronous motor field orientation control system, so as to realize the synchronous iterative execution of parameter identification and motor control.

7. The method for online parameter identification of a surface-mounted permanent magnet synchronous motor according to claim 1, characterized in that, In step 6), the method for determining the convergence of the current error signal is as follows: set the d-axis error threshold and the q-axis error threshold. When the absolute value of the d-axis current error is continuously less than the d-axis error threshold and the absolute value of the q-axis current error is continuously less than the q-axis error threshold in multiple consecutive control cycles, the current error signal is determined to converge, and the identification parameters obtained in the current iteration are output as the effective electrical parameters of the motor.

Images