A permanent magnet synchronous motor parameter identification method based on inverter nonlinearity compensation

Abstract

The present application relates to the technical field of motor control, and particularly relates to a permanent magnet synchronous motor parameter identification method based on inverter nonlinear compensation, comprising: obtaining comprehensive nonlinear error for a motor to be identified and compensating three-phase voltage to obtain compensated dq-axis voltage, establishing a dq-axis discrete identification model of recursive least square identification format with a forgetting factor; injecting excitation into the d-axis in the offline stage and obtaining offline identification results; and identifying motor parameters based on the dq-axis discrete identification model in the online state. In view of the problem that the motor control system in the prior art is prone to introducing more nonlinear errors due to device problems, leading to inaccurate identification results, the comprehensive nonlinear error of the motor is measured and compensated in advance to obtain compensated dq-axis voltage, and then offline and online stage identification is performed, thereby improving the accuracy of identification, avoiding the problem of motor parameter deviation caused by nonlinear error, and the problem of inaccurate identification and control.

Description

Technical Field

[0001] This invention relates to the field of motor control technology, and specifically to a method for parameter identification of permanent magnet synchronous motors based on inverter nonlinear compensation. Background Technology

[0002] In today's industrial and automation fields, permanent magnet synchronous motors (PMSMs) have become one of the most favored drive solutions due to their superior performance, including high efficiency, high power factor, low noise, and excellent dynamic response characteristics. These motors are widely used in electric vehicles, industrial robotic arms, aerospace, and home appliances due to their high reliability and low maintenance requirements. With continuous technological development and the expansion of application areas, the control requirements for PMSMs are becoming increasingly stringent. These advanced control algorithms typically rely on mathematical models of the motor for implementation. The accuracy of model parameters such as the motor's resistance, inductance, and flux linkage directly affects the performance of the control algorithm, thus determining the precision and efficiency of motor control.

[0003] There are already many technical solutions for this problem in the existing technology.

[0004] For example, patent application CN202110931783.5 discloses an online parameter identification method for permanent magnet synchronous motors based on inverter dead-time compensation. The method includes the following steps: Based on the vector control of the permanent magnet synchronous motor, the nonlinearity of the inverter is studied, considering the influence of the inverter dead time on parameter identification and performing online dead-time compensation for the error voltage caused by the dead time; based on the dead-time compensation, online identification is performed using the least squares method with a fast execution rate (current loop execution rate) to identify the d-axis inductance and q-axis inductance; based on the dead-time compensation, the d-axis inductance and q-axis inductance, stator resistance, and rotor flux linkage are identified using the speed loop execution rate. By employing two least squares methods with different execution rates, the simultaneous online identification of the four parameters of the motor—stator resistance, d-axis inductance, q-axis inductance, and rotor flux linkage—is achieved, solving the underrank problem of the voltage equation and improving the identification accuracy.

[0005] For example, patent application CN202211219647.4 discloses a position-free model predictive control system for automatically correcting parameter offsets in permanent magnet synchronous motors. This system includes a dual-loop model predictive control system consisting of a speed loop with deadbeat model predictive control and a current loop with finite set model predictive control; position and load torque observers constructed based on an improved differential disturbance-free observation principle; and parameter identification achieved through a recursive least squares model based on a forgetting factor. This system enables position-free model predictive control of permanent magnet synchronous motors, and when parameter offsets occur, precise motor control is achieved through correction via the identification loop. The scheme considers the impact of real-time parameter changes on the model predictive controller and position observer, ensuring the smooth and robust operation of the entire system.

[0006] However, in practical implementation, the inventors discovered that in permanent magnet synchronous motor drive systems, the voltage command output by the controller is not entirely equivalent to the actual voltage applied to the motor terminals. Nonlinear factors such as inverter dead time, power device on-state voltage drop, switching delay, and parasitic parameters introduce a comprehensive error voltage related to the phase current. If the uncompensated voltage command is directly used as the input to the identification model during parameter identification, the inverter voltage error can easily be mistaken for changes in motor resistance, inductance, or flux linkage. Existing online observer-based compensation methods often rely on motor model parameters, which can easily become coupled with the parameters to be identified; compensation for dead time alone is insufficient to cover the comprehensive nonlinear error caused by the combined effects of on-state voltage drop, switching delay, device differences, and parasitic parameters. Summary of the Invention

[0007] To address the aforementioned problems in existing technologies, a method for parameter identification of permanent magnet synchronous motors based on inverter nonlinear compensation is provided.

[0008] The specific technical solution is as follows: A parameter identification method for permanent magnet synchronous motor based on inverter nonlinear compensation, comprising: Step S1: Obtaining the comprehensive nonlinear error of the motor to be identified and compensating the three-phase voltage to obtain the compensated dq-axis voltage; Step S2: Establishing a discrete identification model of the dq-axis based on the compensated dq-axis voltage; The dq-axis discrete identification model is a recursive least squares identification format with a forgetting factor; Step S3: In the offline stage, inject excitation into the d-axis and perform identification to obtain offline identification results; Step S4: Based on the offline identification results, tune the controller parameters of the motor to be identified and start the motor to be identified; Step S5: In the online state, perform parameter identification based on the dq-axis discrete identification model to obtain motor parameters.

[0009] On the other hand, step S1 includes: step S11: generating multiple non-uniform calibration nodes for the motor to be identified and measuring the comprehensive nonlinear error; step S12: establishing a voltage compensation table based on the comprehensive nonlinear error; step S13: constructing an estimated dq-axis voltage for the offline target current predetermined in the offline identification stage; step S14: looking up the voltage compensation table based on the offline target current and compensating the estimated dq-axis voltage to obtain the compensated dq-axis voltage.

[0010] On the other hand, step S11 includes: step S111: configuring the maximum calibration current, minimum calibration current, and number of unidirectional calibration nodes for the motor to be identified as calibration configuration information; step S112: configuring multiple non-uniform calibration nodes based on the configuration calibration information; step S113: dragging the rotor of the motor to be identified to a preset angle, and inputting d-axis calibration current commands to each non-uniform calibration node to collect the steady-state d-axis calibration current and the corresponding d-axis calibration voltage; step S114: calculating the comprehensive nonlinear error based on the d-axis calibration current and the d-axis calibration voltage.

[0011] On the other hand, in step S112, a low current density region threshold is determined between the maximum calibration current and the minimum calibration current; for the portion where the calibration current is less than the low current density region threshold but greater than the minimum calibration current, logarithmic nodes of a first density are used to generate the non-uniform calibration nodes; for the portion where the calibration current is greater than the low current density region threshold but less than the maximum calibration current, logarithmic nodes of a second density are used to generate the non-uniform calibration nodes; the second density is less than the first density.

[0012] On the other hand, in both steps S3 and S5, the sampled values ​​are compensated based on the voltage compensation table before identification.

[0013] On the other hand, step S2 includes: step S21: establishing a discretized motor dq-axis model for the compensated dq-axis voltage; step S22: converting the discretized motor dq-axis model into a least squares identification model with a least squares identification format; step S23: optimizing the least squares identification model to obtain the dq-axis discrete identification model.

[0014] On the other hand, the discretized motor dq-axis model includes: ; In the formula, The d-axis voltage is after nonlinear compensation by the inverter. These are the q-axis voltages after nonlinear compensation by the inverter. For d-axis current, For q-axis current, For rotational speed, Sampling time, Indicates the sampling time. 、 、 、 These are the motor parameters; The least squares identification model includes: ; In the least squares identification model Input for the model; For model output, These are the motor parameters; The dq-axis discrete identification model includes: ; In the formula, For the gain vector, Let covariance matrix be the variance matrix. It is the identity matrix. Forgetting factor, .

[0015] On the other hand, step S3 includes: step S31: setting a first voltage excitation on the d-axis of the motor to be identified, and measuring the corresponding first d-axis current data; the first voltage excitation is an m-sequence excitation; step S32: calculating the offline identification result based on the first d-axis current data and the dq-axis discrete identification model.

[0016] On the other hand, step S4 includes: step S41: starting the motor to be identified based on the controller parameters; step S42: setting a second voltage excitation on the d-axis of the motor to be identified and measuring the corresponding second d-axis current data; the second voltage excitation is an m-sequence excitation; step S43: calculating the online identification result based on the second d-axis current data combined with the dq-axis discrete identification model.

[0017] A storage medium includes computer instructions, which, when executed by a computer device, perform the aforementioned method for identifying parameters of a permanent magnet synchronous motor.

[0018] The above technical solution has the following advantages or beneficial effects: In view of the problem that the motor control system in the prior art is prone to introducing a lot of nonlinear errors due to the device problem, resulting in inaccurate identification results, by measuring the comprehensive nonlinear error of the motor in advance and compensating for it to obtain the compensated dq axis voltage, and then performing identification in offline and online stages, the accuracy of identification is improved, and the problem of motor parameter deviation and inaccurate identification and control caused by nonlinear error is avoided. Attached Figure Description

[0019] Embodiments of the invention will be described more fully with reference to the accompanying drawings. However, the drawings are for illustration and explanation only and do not constitute a limitation on the scope of the invention.

[0020] Figure 1 This is an overall schematic diagram of an embodiment of the present invention; Figure 2 This is a schematic diagram of step S1 in an embodiment of the present invention; Figure 3 This is a schematic diagram of step S11 in an embodiment of the present invention; Figure 4 This is a schematic diagram of step S2 in an embodiment of the present invention; Figure 5 This is a schematic diagram of step S3 in an embodiment of the present invention; Figure 6 This is a schematic diagram of step S4 in an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0023] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0024] This invention includes: a method for parameter identification of permanent magnet synchronous motors based on inverter nonlinear compensation, such as... Figure 1As shown, the process includes: Step S1: Obtaining the comprehensive nonlinear error of the motor to be identified and compensating the three-phase voltage to obtain the compensated dq-axis voltage; Step S2: Establishing a dq-axis discrete identification model based on the compensated dq-axis voltage; the dq-axis discrete identification model is a recursive least squares identification format with a forgetting factor; Step S3: Injecting excitation into the d-axis and performing identification in the offline stage to obtain the offline identification result; Step S4: Tuning the controller parameters of the motor to be identified based on the offline identification result and starting the motor to be identified; Step S5: Performing parameter identification based on the dq-axis discrete identification model in the online state to obtain the motor parameters.

[0025] Specifically, to address the problem that existing motor control systems are prone to introducing numerous nonlinear errors due to device issues, leading to inaccurate identification results, this paper proposes a method that pre-measures and compensates for the comprehensive nonlinear errors in the motor to obtain the compensated dq-axis voltage. Then, identification is performed sequentially in offline and online stages, improving the accuracy of identification and avoiding the problems of motor parameter deviation and inaccurate identification and control caused by nonlinear errors.

[0026] In practical implementation, the aforementioned permanent magnet synchronous motor parameter identification method is mainly configured as a software implementation in computer equipment, such as the controller of a motor drive system. The motor drive system consists of a controller, a three-phase inverter, a permanent magnet synchronous motor, a current sampling unit, a speed or position detection unit, and a DC bus voltage sampling unit. The controller can be a digital signal processor, microcontroller, programmable logic device, or other computer equipment with computing capabilities. The controller outputs a three-phase voltage command, which, after space vector pulse width modulation (SPWM), drives the three-phase inverter. The inverter converts the DC bus voltage into a three-phase AC voltage and applies it to the permanent magnet synchronous motor. Typically, the controller adopts a dual closed-loop structure with an outer speed loop and an inner current loop. The speed loop generates a q-axis current command based on the speed command and the measured or estimated speed; the current loop generates a dq-axis voltage command based on the dq-axis current command and the dq-axis feedback current; the dq-axis voltage command, after inverse Park transformation, SVPWM, and inverter processing, is applied to the motor. The software program corresponding to the aforementioned permanent magnet synchronous motor parameter identification method is configured in the controller, which reads the compensated dq-axis voltage, dq-axis current, and electric angular velocity, and outputs the direct-axis inductance. quadrature axis inductance 、 Stator resistance 、 Permanent magnet flux These parameters serve as identification parameters, enabling the controller to adjust the control program.

[0027] During the processing, the inverter's comprehensive nonlinear error is first calibrated offline. The controller generates multiple non-uniform calibration nodes according to the preset calibration configuration, applies the corresponding calibration current in sequence, collects the steady-state current and corresponding voltage, calculates the mapping relationship between phase current and error voltage, and forms a voltage compensation table.

[0028] This voltage compensation table is used during offline and online identification processes. The controller queries the voltage compensation table based on the current three-phase current or the predetermined target current for the offline phase to obtain the three-phase error voltage, then transforms it to the dq coordinate system and compensates for the estimated dq-axis voltage. Therefore, the d-axis voltage used in the identification model... shaft voltage It is closer to the actual voltage applied to the motor terminals, rather than just the voltage amount containing the ideal command of the controller, thus avoiding voltage errors caused by nonlinear factors such as inverter dead time, power device conduction voltage drop, switching delay, and parasitic parameters.

[0029] Subsequently, based on the compensated dq-axis voltage, dq-axis current, and electric angular velocity, the controller discretizes the dq-axis voltage equation of the permanent magnet synchronous motor and transforms it into a standard least squares form. Then, a recursive least squares update formula with a forgetting factor is introduced to form a dq-axis discrete identification model, concentrating the parameters to be identified into a parameter vector. This allows new data from each sampling period to be directly used for parameter correction; the recursive form avoids repeatedly solving large-scale matrices; and the forgetting factor enables the algorithm to track parameter drift caused by temperature changes, magnetic saturation, or changes in the operating point.

[0030] After completing the aforementioned preparations, an offline identification phase is performed. With the motor stationary or the rotor fixed, a first voltage excitation is injected only along the d-axis, and the corresponding first d-axis current data is measured. The first voltage excitation uses an m-sequence excitation. The controller utilizes the compensated d-axis voltage and current responses, and through the dq-axis discrete identification model or its offline simplified model, obtains at least the direct-axis inductance... and stator resistance The offline identification results. For surface-mount motors, it is possible to directly... As initial control parameters; for built-in motors, the direct-axis inductance can still be identified using the offline d-axis equation. and stator resistance quadrature axis inductance The initial value can be a nominal value or a conservative value, and it is further updated during the online phase.

[0031] Subsequently, based on the direct-axis inductance obtained offline... and stator resistance The controller calculates the proportional-integral parameters of the current loop, the initial parameters of the speed loop, or the voltage feedforward parameters, and starts the motor with these controller parameters to match the initial closed-loop control parameters with the actual motor. Then, the motor is allowed to enter stable closed-loop operation from a stationary state.

[0032] After the motor achieves stable closed-loop operation, a second voltage excitation is superimposed on the d-axis. This second voltage excitation also employs an m-sequence excitation. In each sampling cycle, the controller compensates for the voltage by looking up a table based on the real-time current. The compensated dq-axis voltage, dq-axis current, and electrical angular velocity are then input into the same recursive least-squares framework, synchronously updating the direct-axis inductance. quadrature axis inductance 、 Stator resistance 、 Permanent magnet flux To further optimize the motor's operating accuracy.

[0033] In one embodiment, such as Figure 2 As shown, step S1 includes: step S11: generating multiple non-uniform calibration nodes for the motor to be identified and measuring the comprehensive nonlinear error; step S12: establishing a voltage compensation table based on the comprehensive nonlinear error; step S13: constructing an estimated dq-axis voltage for the offline target current predetermined in the offline identification stage; step S14: looking up the voltage compensation table based on the offline target current and compensating the estimated dq-axis voltage to obtain the compensated dq-axis voltage.

[0034] Among them, such as Figure 3 As shown, step S11 includes: Step S111: Configure the maximum calibration current, minimum calibration current, and number of unidirectional calibration nodes of the motor to be identified as calibration configuration information; Step S112: Configure multiple non-uniform calibration nodes based on the configuration calibration information; Step S113: Drag the rotor of the motor to be identified to a preset angle, and input the d-axis calibration current command to each non-uniform calibration node to collect the steady-state d-axis calibration current and the corresponding d-axis calibration voltage; Step S114: Calculate the comprehensive nonlinear error based on the d-axis calibration current and d-axis calibration voltage.

[0035] Specifically, in order to achieve a better compensation effect, in this embodiment, in order to eliminate nonlinear errors, the maximum and minimum rated currents of the motor to be identified are first determined according to the current range in the motor nameplate, and multiple non-uniform calibration nodes are divided within this range according to a preset method, wherein the non-uniform calibration nodes cover the zero-crossing point and the area near the zero point of the motor.

[0036] Then, for non-uniform calibration nodes, the rotor of the motor to be identified is dragged and fixed to a preset electrical angle, so that the SVPWM output vector remains fixed relative to the rotor position. The q-axis current is then set. The d-axis calibration current commands corresponding to each non-uniform calibration node are sequentially input to the d-axis; after maintaining a sufficient time at each node until the current enters a steady state, multiple sets of d-axis calibration currents and corresponding d-axis calibration voltages are collected, the sampling results are averaged, and the comprehensive nonlinear error is further calculated.

[0037] A voltage compensation table was then established to reflect the overall nonlinear error.

[0038] During the offline phase, the predetermined offline target current is: , The controller can construct an estimated dq-axis voltage, including the estimated d-axis voltage, based on the motor's nominal parameters, the current loop output from the previous sampling period, or a simplified d-axis voltage equation. Predict the q-axis voltage The controller converts the offline target current into a three-phase target current, looks up the three-phase error voltage in the voltage compensation table, and then transforms it back to the dq coordinate system to obtain the d-axis error voltage. and q-axis error voltage .

[0039] According to the above error sign convention, the compensated dq-axis voltage used for identification is:

[0040] During the online identification phase, the controller samples the three-phase current in real time. Consult the voltage compensation table for each step. When the real-time current is between two adjacent calibration nodes, linear interpolation, piecewise linear interpolation, or spline interpolation can be used to obtain the error voltage for compensation.

[0041] In configuring non-uniform calibration nodes, the range covers both positive and negative current directions, and includes zero current points or several small current points near zero current.

[0042] In one generation method, let the nth positive calibration node... satisfy: ; in ,make and , For the maximum rated current, This is the minimum rated current. This represents the total number of nodes.

[0043] Subsequently, negative calibration nodes are symmetrically generated, and the zero-current point and several small-current nodes near the zero current are added to the calibration sequence.

[0044] In another generation method, in step S112, a low current density region threshold is determined between the maximum calibration current and the minimum calibration current; for the portion where the calibration current is less than the low current density region threshold but greater than the minimum calibration current, logarithmic nodes of a first density are used to generate non-uniform calibration nodes; for the portion where the calibration current is greater than the low current density region threshold but less than the maximum calibration current, logarithmic nodes of a second density are used to generate non-uniform calibration nodes; the second density is less than the first density.

[0045] In one embodiment, such as Figure 4 As shown, step S2 includes: step S21: establishing a discretized motor dq-axis model for compensating the dq-axis voltage; step S22: converting the discretized motor dq-axis model into a least squares identification model with a least squares identification format; step S23: optimizing the least squares identification model to obtain a discrete identification model for the dq-axis.

[0046] The discretized motor dq-axis model includes: ; In the formula, The d-axis voltage is after nonlinear compensation by the inverter. These are the q-axis voltages after nonlinear compensation by the inverter. For d-axis current, For q-axis current, For rotational speed, Sampling time, Indicates the sampling time. 、 、 、 These are the motor parameters; Least squares identification models include: ; In the least squares identification model, Input for the model; For model output, These are the motor parameters; The discrete identification model for the dq axis includes: ; In the formula, For the gain vector, Let covariance matrix be the variance matrix. For the current, Forgetting factor, .

[0047] Specifically, to facilitate subsequent identification, in this embodiment, under the rotor magnetic field orientation dq coordinate system, the continuous voltage equation of the permanent magnet synchronous motor is first established: ; in, The voltage value must be the one after nonlinear compensation by the inverter. The current derivative is discretized using the backward Euler method, i.e.:

[0048] The discretized motor dq-axis model is obtained: ; in, Obtained from the three-phase current sampling values ​​through coordinate transformation. Obtained by the position or speed detection unit. The voltage input in the above model has already undergone pre-compensation in step S1, so it mainly reflects the resistance voltage drop, inductance voltage drop, rotational back EMF, and permanent magnet back EMF of the motor body.

[0049] Then, the discretized motor dq-axis model is transcribed as follows: ; To make the model input matrix Corresponding to each item in the parameter vector, this embodiment adopts the following arrangement:

[0050] Accordingly:

[0051] The model output is:

[0052] It should be noted that the order of the parameter vectors does not change the essence of the identification model. If the parameter vectors are written as... Only the model input matrix needs to be exchanged synchronously. The first two columns yielded It is completely consistent with the model mentioned above.

[0053] Then, the least squares criterion was used as the objective function to be optimized. ; ; Then in the At each sampling time point, the weighted least squares objective function is used:

[0054] Finally, we obtain the calculation formula for recursive least squares with forgetting factor.

[0055]

[0056]

[0057] in, For the first Parameter estimates after sampling; This is the gain matrix; Let covariance matrix be the variance matrix. It is the identity matrix. Forgetting factor, .when Historical data and current data have the same weight; when In this case, earlier data has lower weight, thus making parameter estimation more focused on the current operating state.

[0058] As an optional implementation method, initial parameters Offline identification results or nominal motor parameters can be used. To prevent the parameters from ceasing to update due to an excessively small covariance matrix during long-term operation, a covariance reset can be performed when the minimum eigenvalue of the covariance matrix falls below a preset threshold.

[0059] Among them, prediction error This represents the deviation between the current voltage measurement and the predicted parameter value at the previous moment; the gain matrix. This determines the magnitude of the correction for each parameter based on the deviation; covariance matrix. It reflects the uncertainty of parameter estimation. This structure only requires the current sampled data and the state variables from the previous time step, making it suitable for real-time execution within the interrupt cycle of the motor controller.

[0060] In one embodiment, such as Figure 5 As shown, step S3 includes: step S31: setting a first voltage excitation on the d-axis of the motor to be identified, and measuring the corresponding first d-axis current data; the first voltage excitation is an m-sequence excitation; step S32: calculating the offline identification result based on the first d-axis current data combined with the dq-axis discrete identification model.

[0061] Based on the aforementioned model, during the offline identification phase, the motor can be kept stationary, and the rotor fixed at a preset electrical angle. The controller sets the q-axis identification current. The first voltage excitation is injected only along the d-axis. This first voltage excitation is generated from an m-sequence, for example, mapping the "0"s and "1"s of the m-sequence to... and ,in The excitation amplitude is less than the inverter's allowable output voltage and sufficient to produce an observable current response.

[0062] Set under static conditions And with the q-axis current remaining zero or approximately zero, the least squares identification formula for the offline case is rewritten as follows:

[0063] in: ; ;

[0064] The controller acquires the first d-axis current data simultaneously with the injection of the first voltage excitation. In each sampling cycle, it first queries the voltage compensation table based on the predetermined offline target current to compensate for the estimated d-axis voltage, and then calculates the compensated d-axis voltage. Current d-axis current and the previous sampling time Substituting into the above offline least squares scheme, the offline identified direct-axis inductance is updated using the recursive least squares formula with a forgetting factor. Offline identification of stator resistance .

[0065] The Rs and Ld values ​​obtained from offline identification are used for initial closed-loop controller parameter design, such as calculating the proportional-integral parameters of the current loop, the initial parameters of the speed loop, or the voltage feedforward parameters, enabling the motor to enter stable closed-loop operation. Therefore, offline identification not only provides initial numerical values ​​for online identification but also provides reliable preconditions for closed-loop operation.

[0066] In one embodiment, such as Figure 6 As shown, step S4 includes: step S41: start the motor to be identified based on the controller parameters; step S42: set the second voltage excitation on the d-axis of the motor to be identified and measure the corresponding second d-axis current data; the second voltage excitation is an m-sequence excitation; step S43: calculate the online identification result based on the second d-axis current data and the dq-axis discrete identification model.

[0067] Specifically, offline identification yields offline identification of direct-axis inductors. Offline identification of stator resistance Then, the controller uses this result to tune the initial closed-loop controller parameters. This process can be implemented with reference to existing technologies.

[0068] After the controller completes its tuning, it starts the motor to be identified. During startup, the dq axis current, speed error, and DC bus voltage are monitored. When the current does not exceed the limit, the speed error remains within the preset range for a preset time, and the motor does not experience step loss or continuous oscillation, the motor is determined to have entered a closed-loop stable operating state, and then online identification is performed.

[0069] The controller superimposes a second voltage excitation on the d-axis. This second voltage excitation uses an m-sequence excitation with an amplitude set to a small voltage sufficient to excite identifiable parameters without significantly affecting load operation. To avoid causing significant disturbances to normal torque output, the second voltage excitation is still superimposed on the d-axis, rather than applying a large-amplitude excitation directly to the q-axis.

[0070] The initialization parameters for online identification can be:

[0071] in, , The offline identification results are for offline identification of direct-axis inductors. Offline identification of stator resistance ; Suitable for surface-mount motors In built-in motors, a nominal value can be used; The values ​​can be taken from the motor nameplate, factory parameters, or initial values ​​based on experience. Initialization parameters.

[0072] The following process is executed for each online sampling period: First, the three-phase current is collected, and then the direct-axis current is obtained through coordinate transformation. and cross-axis current Obtaining electric angular velocity ; Obtain the estimated or reconstructed three-phase voltage.

[0073] Then, based on the real-time three-phase current, the voltage compensation table is consulted and interpolated to obtain the three-phase error voltage; the three-phase error voltage is then transformed to the dq coordinate system.

[0074] Subsequently, the estimated dq-axis voltage was compensated to obtain the compensated d-axis voltage at time k. Compensation of q-axis voltage at time k .

[0075] Then, construct the model input matrix based on the data from the current and previous sampling periods. With Discretized dq-axis Model .

[0076] Then calculate the gain matrix sequentially. Parameter estimates Covariance Matrix .

[0077] Finally, the online identification results are subjected to amplitude limiting, low-pass filtering, or moving average, and output or used for controller parameter correction after the stability condition is met.

[0078] Since the second voltage excitation acts on the d-axis, and the motor is rotating, the discrete model includes both the d-axis current variation term and... , and Equal rotational coupling terms. Therefore, the online phase can be updated synchronously within the same recursive process. The parameter update results can be periodically written back to the controller to correct the current loop PI parameters, voltage feedforward terms, and flux linkage-related control terms.

[0079] A storage medium includes computer instructions, which, when executed by a computer device, perform the aforementioned method for identifying parameters of a permanent magnet synchronous motor.

[0080] The above are merely preferred embodiments of the present invention and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should recognize that any equivalent substitutions and obvious changes made based on the description and illustrations of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for parameter identification of a permanent magnet synchronous motor based on inverter nonlinear compensation, characterized in that, include: Step S1: Obtain the comprehensive nonlinear error of the motor to be identified and compensate the three-phase voltage to obtain the compensated dq axis voltage; Step S2: Establish a discrete identification model for the dq axis based on the compensated dq axis voltage; the discrete identification model for the dq axis is a recursive least squares identification format with a forgetting factor; Step S3: Inject excitation into the d-axis and perform identification in the offline stage to obtain offline identification results; Step S4: Based on the offline identification results, tune the controller parameters of the motor to be identified and start the motor to be identified; Step S5: In the online state, perform parameter identification based on the discrete identification model for the dq axis to obtain motor parameters.

2. The method for identifying parameters of a permanent magnet synchronous motor according to claim 1, characterized in that, Step S1 includes: Step S11: Generating multiple non-uniform calibration nodes for the motor to be identified and measuring the comprehensive nonlinear error; Step S12: Establishing a voltage compensation table based on the comprehensive nonlinear error; Step S13: Constructing an estimated dq-axis voltage for the offline target current predetermined in the offline identification stage; Step S14: Looking up the voltage compensation table based on the offline target current and compensating the estimated dq-axis voltage to obtain the compensated dq-axis voltage.

3. The method for identifying parameters of a permanent magnet synchronous motor according to claim 2, characterized in that, Step S11 includes: Step S111: Configure the maximum calibration current, minimum calibration current, and number of unidirectional calibration nodes for the motor to be identified as calibration configuration information; Step S112: Configure multiple non-uniform calibration nodes based on the configured calibration information; Step S113: Drag the rotor of the motor to be identified to a preset angle, and input d-axis calibration current commands to each non-uniform calibration node to collect the steady-state d-axis calibration current and the corresponding d-axis calibration voltage; Step S114: Calculate the comprehensive nonlinear error based on the d-axis calibration current and the d-axis calibration voltage.

4. The method for identifying parameters of a permanent magnet synchronous motor according to claim 3, characterized in that, In step S112, a low current density region threshold is determined between the maximum calibration current and the minimum calibration current; for the portion where the calibration current is less than the low current density region threshold but greater than the minimum calibration current, logarithmic nodes of a first density are used to generate the non-uniform calibration nodes; for the portion where the calibration current is greater than the low current density region threshold but less than the maximum calibration current, logarithmic nodes of a second density are used to generate the non-uniform calibration nodes; the second density is less than the first density.

5. The method for identifying parameters of a permanent magnet synchronous motor according to claim 2, characterized in that, In both steps S3 and S5, the sampled values ​​are compensated based on the voltage compensation table before identification.

6. The method for identifying parameters of a permanent magnet synchronous motor according to claim 1, characterized in that, Step S2 includes: Step S21: Establishing a discretized motor dq-axis model for the compensated dq-axis voltage; Step S22: Converting the discretized motor dq-axis model into a least-squares identification model with a least-squares identification format; Step S23: Optimizing the least-squares identification model to obtain the dq-axis discrete identification model.

7. The method for identifying parameters of a permanent magnet synchronous motor according to claim 6, characterized in that, The discretized motor dq-axis model includes: ； In the formula, The d-axis voltage is after nonlinear compensation by the inverter. These are the q-axis voltages after nonlinear compensation by the inverter. For d-axis current, For q-axis current, For rotational speed, Sampling time, Indicates the sampling time. , , , These are the motor parameters; The least squares identification model includes: ； In the least squares identification model Input for the model; For model output, These are the motor parameters; The dq-axis discrete identification model includes: ； In the formula, For the gain vector, Let covariance matrix be the variance matrix. It is the identity matrix. Forgetting factor, .

8. The method for identifying parameters of a permanent magnet synchronous motor according to claim 1, characterized in that, Step S3 includes: Step S31: Setting a first voltage excitation on the d-axis of the motor to be identified, and measuring the corresponding first d-axis current data; the first voltage excitation is an m-sequence excitation; Step S32: Calculating the offline identification result based on the first d-axis current data and the dq-axis discrete identification model.

9. The method for identifying parameters of a permanent magnet synchronous motor according to claim 1, characterized in that, Step S4 includes: Step S41: Start the motor to be identified based on the controller parameters; Step S42: Set a second voltage excitation on the d-axis of the motor to be identified and measure the corresponding second d-axis current data; the second voltage excitation is an m-sequence excitation; Step S43: Calculate the online identification result based on the second d-axis current data and the dq-axis discrete identification model.

10. A storage medium comprising computer instructions, characterized in that, When the computer device executes the computer instructions, it performs the permanent magnet synchronous motor parameter identification method as described in any one of claims 1-9.

Images