Motor control methods, devices and computer program products

By constructing an algebraic demodulation method for rotor flux observations and virtual disturbance signals, the problem of rotor position detection in permanent magnet synchronous motors at zero and low speeds was solved, achieving stable sensorless control and improving the dynamic performance and reliability of the motor.

CN123122370APending Publication Date: 2026-11-17SHENZHEN INST OF ADVANCED TECH CHINESE ACAD OF SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611518835.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing sensorless control methods for permanent magnet synchronous motors struggle to accurately obtain rotor position information at zero or extremely low speeds, and the injection of high-frequency signals leads to noise and losses, affecting control stability and dynamic performance.

Method used

By acquiring stator current and reference voltage, rotor flux observations are constructed. Algebraic demodulation is performed using spatial projection cost function and mutually orthogonal virtual disturbance signals to extract position error gradient and update rotor estimated angle to control motor operation.

Benefits of technology

It eliminates the need for physical high-frequency signal injection, reducing noise and losses, improving rotor position estimation capabilities, and enhancing control stability and reliability under zero-speed and dynamic operating conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123122370A_ABST
    Figure CN123122370A_ABST
Patent Text Reader

Abstract

This application relates to the field of motor control technology, and provides a motor control method, device, and computer program product. The motor control method includes acquiring the stator current and reference voltage of the motor in a two-phase stationary coordinate system to obtain rotor flux linkage observations; constructing a spatial projection cost function based on the rotor flux linkage observations and the estimated rotor angle, and generating mutually orthogonal first and second virtual disturbance signals in the digital control domain to obtain corresponding cost responses and differential responses, respectively; performing algebraic demodulation based on the two differential responses and the corresponding virtual disturbance signals to obtain the position error gradient, and updating the estimated rotor angle based on the position error gradient to control motor operation. This technical solution does not require injecting physical high-frequency signals into the motor, and can cancel carrier components through algebraic demodulation, reduce electromagnetic noise and phase lag caused by low-pass filtering, and improve the dynamic performance of rotor position estimation and motor closed-loop control under zero-speed and low-speed conditions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of motor control technology, and in particular to a motor control method, device and computer program product. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs) possess advantages such as high power density, high operating efficiency, and good dynamic response performance, making them widely used in industrial automation, electric vehicles, and home appliances. To achieve field-oriented control of PMSMs, accurate rotor position and speed information is typically required. Existing technologies use mechanical position sensors to detect rotor position; however, mechanical position sensors increase the size and cost of the motor control system and suffer from reliability degradation in environments with high temperature, high vibration, and heavy dust. Therefore, sensorless control technology has become widely adopted.

[0003] Existing sensorless control methods for permanent magnet synchronous motors typically include observation methods based on back electromotive force (EMF) or flux linkage, and high-frequency signal injection methods. The observation method based on back EMF or flux linkage acquires the motor's voltage and current, uses a back EMF observer or flux linkage observer to obtain the back EMF or flux linkage, and further extracts rotor position information through methods such as phase-locked loops. However, the amplitude of the back EMF is related to the motor speed. When the motor is at zero speed or extremely low speed, the back EMF is small or even close to zero, reducing the observability of the system and making it difficult to accurately obtain rotor position information. Therefore, under conditions such as load start-up or sudden load changes, problems such as starting reversal, loss of synchronization, or starting failure are prone to occur.

[0004] To improve rotor position detection performance at zero and low speeds, existing technologies can inject high-frequency voltage signals onto the d-axis or q-axis of the motor and extract rotor position information based on the high-frequency current response. However, this method requires the actual application of a high-frequency excitation signal to the motor via an inverter. The high-frequency current flowing through the stator windings easily generates high-frequency noise, as well as additional iron and copper losses. Furthermore, its position recognition capability typically depends on the salient pole characteristics of the motor, limiting its applicability to motors with weak salient pole effects. In addition, demodulating the high-frequency response signal usually requires using a low-pass filter to extract the position error signal. However, the low-pass filter introduces a certain phase lag, which can lead to a large dynamic deviation between the estimated rotor angle and the actual rotor angle during rapid acceleration / deceleration or when subjected to step load impacts. This can affect the stability and dynamic performance of sensorless motor control. Summary of the Invention

[0005] This application provides a motor control method, apparatus, and computer program product to solve the above-mentioned technical problems.

[0006] Firstly, a motor control method is provided, the motor control method comprising: The stator current, reference voltage, and estimated rotor angle of the motor are obtained in a two-phase stationary coordinate system. The rotor flux linkage observation value of the motor is obtained based on the stator current and the reference voltage. The rotor flux linkage observation value includes a first rotor flux linkage observation value and a second rotor flux linkage observation value. A spatial projection cost function is constructed, and based on the first rotor flux linkage observation value, the second rotor flux linkage observation value, and the estimated rotor angle, the first cost response corresponding to the estimated rotor angle when no virtual disturbance is applied is determined through the spatial projection cost function. Generate a first virtual disturbance signal and a second virtual disturbance signal that are orthogonal to each other; determine a second cost response using the spatial projection cost function and the first virtual disturbance signal; and determine a third cost response using the spatial projection cost function and the second virtual disturbance signal. A first differential response is determined based on the second cost response and the first cost response, and a second differential response is determined based on the third cost response and the first cost response; Algebraic demodulation is performed based on the first differential response, the second differential response, the first virtual disturbance signal, and the second virtual disturbance signal to obtain the position error gradient, and the rotor estimated angle is updated based on the position error gradient to control the operation of the motor.

[0007] Optionally, obtaining the observed rotor flux linkage of the motor based on the stator current and the reference voltage includes: Dynamic back electromotive force is observed based on the stator current, the reference voltage, and the motor parameters of the motor to obtain the observed dynamic back electromotive force value. The rotor flux linkage observation value is obtained by performing hybrid flux linkage observation based on the dynamic back electromotive force observation value and the steady-state current model reference value.

[0008] Optionally, the spatial projection cost function is: ; in, This is the first rotor flux linkage observation value. Estimate the angle for the rotor. To estimate the α-axis direction component corresponding to the angle, This is the second rotor flux linkage observation value. To estimate the β-axis direction component corresponding to the angle.

[0009] Optionally, determining the second cost response using the spatial projection cost function and the first virtual perturbation signal, and determining the third cost response using the spatial projection cost function and the second virtual perturbation signal, includes: The rotor estimated angle is perturbed according to the first virtual disturbance signal to obtain a first disturbance angle, and the rotor estimated angle is perturbed according to the second virtual disturbance signal to obtain a second disturbance angle. The second cost response is determined based on the spatial projection cost function and the first disturbance angle, and the third cost response is determined based on the spatial projection cost function and the second disturbance angle.

[0010] Optionally, the step of performing algebraic demodulation based on the first differential response, the second differential response, the first virtual disturbance signal, and the second virtual disturbance signal to obtain the position error gradient includes: A first demodulation result is determined based on the first differential response and the first virtual disturbance signal, and a second demodulation result is determined based on the second differential response and the second virtual disturbance signal. The first demodulation result and the second demodulation result are added together, and the corresponding carrier components are canceled out according to the orthogonality between the first virtual disturbance signal and the second virtual disturbance signal to obtain the position error gradient.

[0011] Optionally, updating the rotor estimated angle based on the position error gradient includes: The flux linkage amplitude parameter is determined based on the first rotor flux linkage observation value and the second rotor flux linkage observation value; The position error gradient is normalized according to the magnetic flux amplitude parameter to obtain a normalized position error signal; The rotor estimated angle is updated based on the normalized position error signal.

[0012] Optionally, updating the rotor estimated angle based on the normalized position error signal includes: The normalized position error signal is input into a third-order extended state observer, and the state of the normalized position error signal is observed through the third-order extended state observer to obtain the rotor angle state, rotor speed state and rotor acceleration state. The estimated rotor angle is updated based on the rotor angle state.

[0013] Optionally, controlling the operation of the motor includes: Based on the updated rotor estimated angle, Park transformation and inverse Park transformation are performed to perform field-oriented control of the motor; A pulse width modulation signal is generated based on the magnetic field orientation control result to control the inverter to drive the motor.

[0014] A second aspect of the present invention provides a motor control device, the motor control device comprising: The rotor flux linkage observation module is used to acquire the stator current, reference voltage and rotor estimated angle of the motor in a two-phase stationary coordinate system, and to acquire the rotor flux linkage observation value of the motor based on the stator current and the reference voltage. The rotor flux linkage observation value includes a first rotor flux linkage observation value and a second rotor flux linkage observation value. The cost response acquisition module is used to construct a spatial projection cost function, determine a first cost response corresponding to the estimated rotor angle when no virtual disturbance is applied based on the first rotor flux linkage observation value, the second rotor flux linkage observation value, and the estimated rotor angle; generate a first virtual disturbance signal and a second virtual disturbance signal that are mutually orthogonal, determine a second cost response based on the spatial projection cost function and the first virtual disturbance signal, and determine a third cost response based on the spatial projection cost function and the second virtual disturbance signal. The differential response acquisition module is used to determine a first differential response based on the second cost response and the first cost response, and to determine a second differential response based on the third cost response and the first cost response; The position error gradient acquisition module is used to perform algebraic demodulation based on the first differential response, the second differential response, the first virtual disturbance signal, and the second virtual disturbance signal to obtain the position error gradient. The motor control module is used to update the rotor estimated angle according to the position error gradient in order to control the operation of the motor.

[0015] A third aspect of the present invention provides a computer program product, the computer program product storing a computer program, the computer program including program instructions, the program instructions being executed by a computer to cause the computer to perform the method described in the first aspect.

[0016] The technical advantages of this invention are as follows: Rotor flux linkage observations in a two-phase stationary coordinate system are obtained using stator current and reference voltage. A spatial projection cost function is constructed based on the rotor flux linkage observations and the estimated rotor angle. Furthermore, a first orthogonal virtual disturbance signal and a second virtual disturbance signal are generated in the digital control domain. Differential responses are obtained using the cost responses corresponding to the two virtual disturbances and the cost response without virtual disturbances. Algebraic demodulation is then performed on the two differential responses to extract the position error gradient. Therefore, on the one hand, position error information for correcting the estimated rotor angle can be obtained without actually injecting a physical high-frequency voltage signal into the motor, reducing electromagnetic noise and additional losses generated by physical high-frequency signal injection. On the other hand, by utilizing the orthogonality between the two virtual disturbance signals for algebraic demodulation, corresponding carrier components can be canceled during algebraic operations, eliminating the need for a low-pass filter to extract position error information. This reduces phase lag caused by low-pass filtering and improves the dynamic response performance of position error extraction and rotor position tracking. Based on this, the rotor estimated angle is continuously updated according to the obtained position error gradient and the motor operation is controlled. This is beneficial to improve the rotor position estimation capability of the motor under zero speed, extremely low speed and dynamic speed change conditions, realize the direct establishment of closed-loop control of the motor from zero speed, and improve the stability and reliability of the motor during load start-up and dynamic operation. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a motor control method according to Embodiment 1 of the present invention; Figure 2 This is a flowchart illustrating the specific execution process of step S101 in a motor control method according to Embodiment 1 of the present invention. Figure 3 This is a flowchart illustrating the specific execution process of step S103 in a motor control method according to Embodiment 1 of the present invention. Figure 4 This is one of the specific workflow diagrams for executing step S105 in a motor control method according to Embodiment 1 of the present invention; Figure 5 This is the second detailed flowchart of the execution step S105 in a motor control method according to Embodiment 1 of the present invention; Figure 6 This is a flowchart illustrating the specific execution process of step S503 in a motor control method according to Embodiment 1 of the present invention. Figure 7 This is the third detailed flowchart of the execution step S105 in a motor control method according to Embodiment 1 of the present invention; Figure 8 This is a curve showing the amplitude characteristic of a closed-loop third-order extended state observer in a motor control method according to Embodiment 1 of the present invention under different observation bandwidths. Figure 9 This is a frequency response curve of a closed-loop third-order extended state observer in a motor control method according to Embodiment 1 of the present invention under different observation bandwidths; Figure 10 This is a schematic diagram of the experimental results of direct zero-speed closed-loop start of a permanent magnet synchronous motor in a motor control method according to Embodiment 1 of the present invention under 50% rated load conditions and with inductance parameter mismatch. Figure 11 This is a schematic diagram of the experimental results of direct zero-speed closed-loop start of a permanent magnet synchronous motor under 100% full load conditions in a motor control method according to Embodiment 1 of the present invention. Figure 12 In one embodiment of the present invention, a permanent magnet synchronous motor in a motor control method rapidly reverses from +100 r / min to... Schematic diagram of the results of the 100 r / min speed reversal experiment; Figure 13 This is a schematic diagram of the experimental results of a permanent magnet synchronous motor operating stably at extremely low speeds in a motor control method according to Embodiment 1 of the present invention; Figure 14 This is a schematic diagram of the motor control device in Embodiment 2 of the present invention. Detailed Implementation

[0019] 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, not all, of the embodiments of the present invention. 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.

[0020] To illustrate the technical solution of the present invention, specific embodiments are described below.

[0021] Example 1 This embodiment provides a motor control method, which includes: Step S101. Obtain the stator current, reference voltage, and estimated rotor angle of the motor in a two-phase stationary coordinate system, and obtain the observed rotor flux linkage of the motor based on the stator current and reference voltage. The observed rotor flux linkage includes the first rotor flux linkage observed value and the second rotor flux linkage observed value.

[0022] Step S102. Construct a spatial projection cost function, and determine the first cost response corresponding to the rotor estimated angle when no virtual disturbance is applied, based on the first rotor flux observation value, the second rotor flux observation value, and the rotor estimated angle.

[0023] Step S103. Generate a first virtual disturbance signal and a second virtual disturbance signal that are mutually orthogonal, and determine the second cost response by using the spatial projection cost function and the first virtual disturbance signal, and determine the third cost response by using the spatial projection cost function and the second virtual disturbance signal.

[0024] Step S104. Determine the first differential response based on the second cost response and the first cost response, and determine the second differential response based on the third cost response and the first cost response.

[0025] Step S105. Perform algebraic demodulation based on the first differential response, the second differential response, the first virtual disturbance signal, and the second virtual disturbance signal to obtain the position error gradient. Update the rotor estimated angle based on the position error gradient and control the motor operation.

[0026] This embodiment provides a motor control method applicable to sensorless control systems of permanent magnet synchronous motors. By constructing a virtual disturbance within the digital control domain and performing algebraic demodulation on the corresponding cost response, rotor position estimation is achieved. This method eliminates the need to actually apply high-frequency excitation signals to the permanent magnet synchronous motor, reducing high-frequency noise and additional losses caused by physical high-frequency signal injection, and improving rotor position estimation performance under zero-speed or low-speed conditions.

[0027] For step S101, during the operation of the permanent magnet synchronous motor, the motor's operating electrical parameters can be collected in real time, and flux linkage can be observed based on the collected operating electrical parameters. Specifically, the three-phase stator current of the permanent magnet synchronous motor can be converted to the αβ two-phase stationary coordinate system to obtain the stator current in the two-phase stationary coordinate system; the reference voltage can be the voltage command output by the motor control system according to the current control state, or the αβ axis reference voltage calculated from the corresponding voltage command.

[0028] After obtaining the stator current and reference voltage, the flux linkage state of the permanent magnet synchronous motor can be observed using the electrical model of the motor. For example, the dynamic back electromotive force (EMF) observation value can be obtained first based on the stator current, reference voltage, and corresponding motor parameters. Then, a hybrid flux linkage observation can be performed by combining the steady-state current model reference value to obtain the flux linkage observation results in a two-phase stationary coordinate system. The rotor flux linkage observation value obtained in this way can include a first rotor flux linkage observation value along the α-axis direction and a second rotor flux linkage observation value along the β-axis direction. The two flux linkage observation components jointly characterize the flux linkage vector in the two-phase stationary coordinate system and serve as the basis for subsequently constructing the spatial projection cost function and extracting rotor position error information.

[0029] For step S102, a spatial projection relationship can be established using the two flux linkage observation components obtained in step S101 and the current rotor estimated angle. The spatial projection cost function is used to characterize the degree of projection of the flux linkage observation vector in the direction indicated by the rotor estimated angle.

[0030] For example, in the αβ two-phase stationary coordinate system, the corresponding spatial projection result can be obtained by summing the products of the first rotor flux linkage observation value and the cosine of the estimated rotor angle, and the products of the second rotor flux linkage observation value and the sine of the estimated rotor angle. The spatial projection cost function is: ; in, This is the first rotor flux linkage observation value. Estimate the angle for the rotor. To estimate the α-axis direction component corresponding to the angle, This is the second rotor flux linkage observation value. To estimate the β-axis direction component corresponding to the angle.

[0031] The first cost response is obtained without applying a virtual perturbation to the current rotor estimated angle, and therefore can serve as the baseline cost response for the subsequent two perturbation branches. As the direction corresponding to the rotor estimated angle gradually approaches the actual direction of the flux linkage vector, the corresponding spatial projection changes, reaching an extreme value when the two are aligned. Thus, the spatial projection cost function can be used to characterize the positional deviation between the rotor estimated angle and the flux linkage direction.

[0032] Furthermore, there is a correlation between the change in the spatial projection cost function with respect to the rotor estimated angle and the rotor position error. Specifically, the first and second rotor flux linkage observations characterize the components of the flux linkage observation vector along different coordinate axes in the two-phase stationary coordinate system, respectively. The two rotor flux linkage observations jointly determine the amplitude and spatial direction of the flux linkage observation vector. The spatial direction of the flux linkage observation vector characterizes the actual position information of the motor rotor, while the spatial projection cost function characterizes the projection of the flux linkage observation vector onto the direction corresponding to the rotor estimated angle. Therefore, when the rotor estimated angle changes, the projection of the flux linkage observation vector onto the corresponding direction changes accordingly, and this change is related to the angular deviation between the rotor estimated angle and the actual rotor position. Thus, the rate of change of the spatial projection cost function with respect to the rotor estimated angle can characterize the position deviation information between the actual rotor position and the rotor estimated position.

[0033] It should be noted that the actual electrical angle of the rotor does not need to be directly obtained in the above process. The actual position information of the rotor is already included in the flux linkage observation vector formed by the first and second rotor flux linkage observation values. Therefore, the current flux linkage observation vector can be kept unchanged, and the rotor estimated angle used for spatial projection calculation can be changed in the digital control domain to determine the change in the spatial projection cost function with the rotor estimated angle. Since this change depends on the relative positional relationship between the spatial direction of the flux linkage observation vector and the direction corresponding to the rotor estimated angle, information related to the position deviation between the actual rotor position and the rotor estimated position can be extracted from the change in the spatial projection cost function, providing a basis for subsequently determining the position error gradient through virtual perturbation.

[0034] For step S103, two mutually orthogonal virtual disturbance signals can be generated during the software calculation of the motor controller. For example, the two virtual disturbance signals can be a sine signal and a cosine signal with the same virtual disturbance amplitude and virtual disturbance frequency, respectively, i.e.: ,as well as ; in, This indicates the preset virtual disturbance amplitude. This represents the preset virtual perturbation frequency, and t represents time.

[0035] Two virtual disturbance signals are applied to the current estimated rotor angle, resulting in two different disturbance angles in the digital control domain. For example, the first virtual disturbance signal can be superimposed on the current estimated rotor angle to obtain the first disturbance angle, and the second virtual disturbance signal can be superimposed on the current estimated rotor angle to obtain the second disturbance angle. Subsequently, the first and second disturbance angles are substituted into the spatial projection cost function established in step S102, and the second and third cost responses are calculated respectively, while maintaining the corresponding sampling state of the flux linkage observation components.

[0036] The "virtual disturbance" here refers to numerically perturbing the estimated rotor angle within the control algorithm, rather than actually applying high-frequency voltage or current to the permanent magnet synchronous motor windings via the inverter. Therefore, the virtual disturbance itself does not enter the motor's physical power circuit, thus avoiding the additional high-frequency current, noise, and losses caused by traditional physical high-frequency signal injection while obtaining disturbance response information.

[0037] For step S104, the cost response before and after applying the virtual perturbation can be compared to obtain the change in the cost function caused by the two virtual perturbations.

[0038] Specifically, the second cost response can be differentially processed with the first cost response obtained without applying a virtual disturbance to obtain a first differential response; correspondingly, the third cost response can be differentially processed with the first cost response to obtain a second differential response. That is to say, the first differential response reflects the spatial projection change caused by the first virtual disturbance signal acting on the rotor estimated angle, while the second differential response reflects the spatial projection change caused by the second virtual disturbance signal acting on the rotor estimated angle.

[0039] By using the first cost response without virtual perturbation as a common benchmark, the change components caused by the corresponding virtual perturbation can be extracted from the cost responses after the two perturbations. Since the two virtual perturbations are orthogonal to each other, the resulting first and second difference responses can provide a basis for the next step of extracting the changing trend of the spatial projection cost function from the two orthogonal directions.

[0040] For step S105, the first differential response and the second differential response obtained in step S104 can be processed with the corresponding first virtual disturbance signal and the second virtual disturbance signal, respectively, and the two processing results can be combined to achieve algebraic demodulation of the biorthogonal disturbance response.

[0041] For example, the first differential response can be multiplied by the corresponding sinusoidal virtual disturbance signal, and the second differential response can be multiplied by the corresponding cosine virtual disturbance signal. The two product results can then be added together. Since the sinusoidal and cosine virtual disturbance signals are orthogonal, when combining the two responses, the identity relationship between the sum of the squares of the sine and cosine terms can be used to eliminate the virtual high-frequency carrier, thereby extracting the position error gradient related to the rotor position deviation.

[0042] Therefore, compared with the method of extracting position error information from the high-frequency modulation response through a low-pass filter, this embodiment can directly cancel the carrier component through algebraic operations, without relying on a low-pass filter to filter out the high-frequency carrier. This reduces the phase lag introduced by the filtering process and improves the dynamic response speed of the position error information.

[0043] Furthermore, after obtaining the position error gradient, it can be normalized based on the flux linkage observations to obtain a normalized position error signal, thereby reducing the impact of flux linkage amplitude changes on the position error feedback gain. Additionally, by limiting the flux linkage vector amplitude, the flux linkage amplitude parameter used in the normalization operation can be kept non-zero in the zero-velocity state, reducing the possibility of division-zero singularities in the zero-velocity state.

[0044] The normalized position error signal can be further input into the angle tracking stage, for example, into a third-order extended state observer. The third-order extended state observer observes the rotor angle, rotor speed, and rotor acceleration, and outputs an updated rotor estimated angle. The updated rotor estimated angle can be used for Park transform and inverse Park transform to achieve field-oriented control of the permanent magnet synchronous motor. Furthermore, based on the control results, a corresponding pulse width modulation signal is generated to drive the permanent magnet synchronous motor through an inverter.

[0045] Through the above steps S101 to S105, a closed-loop position estimation process of "magnetic flux observation, spatial projection, biorthogonal virtual disturbance, differential response, algebraic demodulation, and position update" can be formed. Position error information can be obtained without actually injecting high-frequency excitation signals into the permanent magnet synchronous motor. The phase lag caused by traditional low-pass filtering demodulation is reduced by the algebraic combination of the biorthogonal virtual disturbance response, which is beneficial to improving the sensorless control performance of the permanent magnet synchronous motor under zero speed, low speed and dynamic operating conditions.

[0046] As one implementation method, such as Figure 2 As shown, step S101, which involves obtaining the observed rotor flux linkage of the motor based on the stator current and reference voltage, includes: Step S201. Observe the dynamic back EMF based on the stator current, reference voltage and motor parameters to obtain the observed dynamic back EMF value.

[0047] Step S202. Perform hybrid flux observation based on the dynamic back electromotive force observation value and the steady-state current model reference value to obtain the rotor flux observation value.

[0048] For step S201, the stator current and reference voltage of the motor in the two-phase stationary coordinate system can be input into the dynamic back EMF observer, and combined with the motor parameters of the permanent magnet synchronous motor, the back EMF generated during the operation of the permanent magnet synchronous motor can be observed. The motor parameters may include stator resistance and stator inductance, the stator current may include the stator current in the α and β two-phase stationary coordinate systems, and the reference voltage may include the reference voltage in the α and β two-phase stationary coordinate systems.

[0049] Specifically, based on the voltage relationship of the permanent magnet synchronous motor in the α and β phase stationary coordinate systems, the dynamic back electromotive force (EMF) related to the rotor flux linkage change can be determined by subtracting the stator resistance voltage drop from the reference voltage, and combining the changes in stator inductance and stator current. The dynamic back EMF observer can observe the α and β axes separately, thereby obtaining observed dynamic back EMF values ​​including the α-axis and β-axis dynamic back EMF components. In other words, this step converts the available stator current and reference voltage into information characterizing the dynamic changes in flux linkage, providing input to the dynamic model side for subsequent flux linkage observations.

[0050] In one specific implementation, the dynamic back electromotive force observation process can be represented as: ; in, This represents the stator current in the α and β two-phase stationary coordinate system. This represents the reference voltage in the α and β two-phase stationary coordinate system. This represents the stator resistance of a permanent magnet synchronous motor. This refers to the stator inductance of a permanent magnet synchronous motor. This represents the observed value of the dynamic back electromotive force.

[0051] Therefore, the dynamic back EMF observer can obtain dynamic back EMF information reflecting changes in electromagnetic state in real time based on the voltage and current at the motor terminals and the motor's own parameters, and provide this dynamic back EMF information to the subsequent hybrid flux observation stage.

[0052] For step S202, the dynamic back EMF observation value obtained in step S201 can be used as information on the dynamic change of flux linkage. At the same time, the steady-state current model reference value is introduced to correct the flux linkage observation process, so that the flux linkage observation can utilize the information provided by the dynamic back EMF model and the steady-state current model.

[0053] Specifically, the dynamic back EMF observations can reflect the dynamic changes in motor flux linkage, and the corresponding flux linkage information can be obtained by integrating the dynamic back EMF observations. However, in actual control processes, current and voltage sampling may have DC bias, and motor parameters may also have certain deviations. If only the dynamic back EMF is relied upon for integration, related errors may accumulate during the integration process. Therefore, a steady-state current model reference value can be further introduced, and a correction value can be formed based on the deviation between the current flux linkage observation value and the steady-state current model reference value. This correction value is then fed back into the flux linkage observation process to constrain and correct the flux linkage observation results.

[0054] In one specific implementation, the hybrid flux observation process can be represented as follows: ; Where, ω c Indicates the crossover frequency. This represents the reference value for the steady-state current model. This represents the observed rotor flux linkage values ​​in the α and β two-phase stationary coordinate systems. Dynamic back electromotive force observation.

[0055] By adjusting the crossover frequency, the roles of dynamic back EMF observation information and steady-state current model reference information in the hybrid flux observation process can be adjusted, allowing the information from the two models to complement each other.

[0056] The steady-state current model reference value can be determined based on the stator current and motor parameters of the permanent magnet synchronous motor. Specifically, based on the current estimated rotor angle, the stator current in the α and β phase stationary coordinate system can be transformed to the d and q phase rotating coordinate system to obtain the d-axis stator current and q-axis stator current. Based on the d-axis stator current, q-axis stator current, and the inductance parameters and permanent magnet flux linkage parameters of the permanent magnet synchronous motor, the d-axis flux linkage reference components and q-axis flux linkage reference components are determined according to the steady-state flux linkage relationship of the permanent magnet synchronous motor. Furthermore, based on the current estimated rotor angle, the d-axis flux linkage reference components and q-axis flux linkage reference components are transformed to the α and β phase stationary coordinate system to obtain the α-axis flux linkage reference components and β-axis flux linkage reference components, and the steady-state current model reference value is determined based on the α-axis flux linkage reference components and β-axis flux linkage reference components.

[0057] Furthermore, the above-mentioned hybrid flux linkage observations can be performed on the α-axis and β-axis respectively to obtain the first rotor flux linkage observation component and the second rotor flux linkage observation component. The two flux linkage observation components together constitute the flux linkage observation vector used by the subsequent spatial projection cost function.

[0058] Furthermore, at zero or extremely low speeds, the weak back electromotive force means that the DC bias in current sampling can significantly impact flux linkage observation results. To address this, the flux linkage vector amplitude obtained from hybrid flux linkage observations can be constrained, keeping it near a preset range corresponding to the permanent magnet flux linkage. This prevents the flux linkage vector from gradually decaying to zero due to observation errors at zero speed. This improves the stability of flux linkage observation results at zero speed and provides a non-zero amplitude basis for subsequent position error normalization based on flux linkage amplitude, thus reducing the possibility of division-by-zero singularities during normalization.

[0059] Through steps S201 and S202, the back electromotive force observation information that reflects the dynamic state of the motor is first obtained by using the stator current, reference voltage and motor parameters. Then, the flux linkage observation process is corrected by using the steady-state current model reference information, so that the dynamic model and the steady-state model cooperate with each other, thereby obtaining a relatively stable flux linkage observation result, and providing a flux linkage information basis for subsequent calculation of spatial projection cost function, extraction of virtual disturbance response and rotor position estimation.

[0060] As one implementation method, such as Figure 3 As shown, step S103, which determines the second cost response using the spatial projection cost function and the first virtual perturbation signal, and determines the third cost response using the spatial projection cost function and the second virtual perturbation signal, includes: Step S301. The rotor estimated angle is disturbed according to the first virtual disturbance signal to obtain the first disturbance angle, and the rotor estimated angle is disturbed according to the second virtual disturbance signal to obtain the second disturbance angle.

[0061] Step S302. Determine the second cost response based on the spatial projection cost function and the first disturbance angle, and determine the third cost response based on the spatial projection cost function and the second disturbance angle.

[0062] For step S301, two mutually orthogonal virtual disturbance signals can be generated in the digital control domain of the motor controller, and the two virtual disturbance signals can be superimposed on the current rotor estimated angle to form two disturbance angles corresponding to different disturbance directions near the current rotor estimated angle.

[0063] For example, the first virtual disturbance signal can be a sinusoidal virtual disturbance signal, and the second virtual disturbance signal can be a cosine virtual disturbance signal. Both can have the same preset virtual disturbance amplitude and preset virtual disturbance frequency. The first virtual disturbance signal and the second virtual disturbance signal can be represented as follows: , ; in, This indicates the preset virtual disturbance amplitude. This indicates the preset virtual perturbation frequency. This indicates time. Because the sine and cosine signals are orthogonal, it is possible to virtually detect the current rotor estimated angle from two mutually orthogonal disturbance directions.

[0064] Specifically, the first virtual disturbance signal can be superimposed with the current estimated rotor angle to obtain the first disturbance angle, and the second virtual disturbance signal can be superimposed with the current estimated rotor angle to obtain the second disturbance angle, i.e.: , : in, This indicates the current estimated rotor angle. Indicates the first disturbance angle. This indicates the second disturbance angle. This indicates the preset virtual disturbance amplitude.

[0065] It should be noted that the above-mentioned disturbance process can be performed only during the software calculation of the controller. That is, the first and second virtual disturbance signals are used to change the rotor estimated angle participating in the spatial projection cost function calculation, without having to convert the two virtual disturbance signals into actual high-frequency voltages or currents and apply them to the permanent magnet synchronous motor. Therefore, a biorthogonal disturbance targeting the rotor estimated angle can be formed in the digital control domain, providing a basis for subsequently obtaining spatial projection changes under different disturbance directions.

[0066] For step S302, after obtaining the first disturbance angle and the second disturbance angle, the rotor estimated angle when no virtual disturbance is applied can be replaced by the two disturbance angles respectively, and substituted into the aforementioned spatial projection cost function to calculate the spatial projection results corresponding to the two disturbance states.

[0067] For example, the spatial projection cost function is expressed as: In this case, the first disturbance angle can be Substituting this into the spatial projection cost function, we obtain the second cost response: Accordingly, the second disturbance angle can be... Substituting the same spatial projection cost function, we obtain the third cost response: .

[0068] in, , These represent the observed values ​​of the first rotor flux linkage and the second rotor flux linkage, respectively. This represents the second cost response corresponding to the first virtual disturbance signal. This represents the third cost response corresponding to the second virtual disturbance signal.

[0069] The first cost response characterizes the spatial projection state when the current rotor estimated angle is not affected by virtual disturbances, while the second and third cost responses characterize the spatial projection states formed after changing the rotor estimated angle along the two orthogonal virtual disturbance directions, respectively. Therefore, by comparing the second cost response with the first cost response, and by comparing the third cost response with the first cost response, the changes in the spatial projection cost function along the two virtual disturbance directions can be obtained, namely the first and second differential responses in subsequent steps.

[0070] Furthermore, within a control cycle, the first cost response, second cost response, and third cost response can be calculated based on the same set of first and second rotor flux linkage observations. Specifically, the current estimated rotor angle is used when calculating the first cost response, while the first and third cost responses are calculated using the first and second disturbance angles formed based on the first and second virtual disturbance signals, respectively. Thus, with the flux linkage observation state remaining the same, the differences between the first, second, and third cost responses primarily stem from the different angles used for spatial projection calculations. Therefore, the differences between the second and first cost responses, and between the third and first cost responses, can characterize the changes in the spatial projection cost function along different virtual disturbance directions near the current estimated rotor angle.

[0071] It should be noted that the first and second virtual disturbance signals act on the rotor estimation angle in the digital control domain to change the angle parameters used in the spatial projection cost function calculation, without changing the voltage command, current command, or pulse width modulation signal actually sent to the motor drive circuit. Therefore, the virtual disturbance does not generate a high-frequency voltage or high-frequency current actually applied to the motor windings, but is used to detect local changes in the spatial projection cost function with respect to the rotor estimation angle in the digital computing domain.

[0072] Through the above steps S301 and S302, without changing the actual voltage and current excitation state of the motor, two mutually orthogonal virtual disturbances can be constructed around the current rotor estimated angle in the digital control domain, and the corresponding spatial projection cost responses can be calculated respectively. This provides information that can reflect the change of the spatial projection cost function with the rotor estimated angle, and provides a data basis for subsequent algebraic demodulation and extraction of position error gradient through the two differential responses.

[0073] As one implementation method, such as Figure 4 As shown, step S105 involves algebraic demodulation based on the first differential response, the second differential response, the first virtual disturbance signal, and the second virtual disturbance signal to obtain the position error gradient, including: Step S401. Determine the first demodulation result based on the first differential response and the first virtual disturbance signal, and determine the second demodulation result based on the second differential response and the second virtual disturbance signal.

[0074] Step S402. Add the first demodulation result and the second demodulation result together, and cancel the corresponding carrier components according to the orthogonality between the first virtual disturbance signal and the second virtual disturbance signal to obtain the position error gradient.

[0075] For step S401, after obtaining the first differential response and the second differential response, synchronous demodulation can be performed using virtual perturbation signals corresponding to each differential response to extract the cost function change information caused by the two virtual perturbations. The first differential response corresponds to the cost response change under the action of the first virtual perturbation signal, and the second differential response corresponds to the cost response change under the action of the second virtual perturbation signal.

[0076] Specifically, the first differential response can be determined by the difference between the second cost response and the first cost response, and the second differential response can be determined by the difference between the third cost response and the first cost response, that is: , ; in, This represents the first cost response when no virtual disturbance is applied to the rotor estimated angle. This represents the second cost response after applying the first virtual perturbation signal. This represents the third cost response after applying the second virtual perturbation signal. and These represent the first and second differential responses, respectively.

[0077] In one specific implementation, the first virtual disturbance signal and the second virtual disturbance signal can be respectively: , ; in, Indicates the virtual disturbance amplitude. Indicates the virtual perturbation frequency. Indicates time.

[0078] The first differential response can be multiplied with the first virtual disturbance signal to obtain the first demodulation result; simultaneously, the second differential response can be multiplied with the second virtual disturbance signal to obtain the second demodulation result. For example, the first and second demodulation results can be expressed as follows: ; .

[0079] Through the above processing, the cost change information related to the first virtual disturbance signal in the first differential response can be mapped to the first demodulation result, and the cost change information related to the second virtual disturbance signal in the second differential response can be mapped to the second demodulation result, thereby forming two demodulation information corresponding to the orthogonal virtual disturbance, providing a basis for subsequent elimination of virtual carriers through algebraic combination.

[0080] For step S402, the first demodulation result and the second demodulation result obtained in step S401 can be algebraically summed to combine the cost function change information carried by the two orthogonal virtual disturbances.

[0081] Specifically, when the first virtual disturbance signal is a sinusoidal virtual disturbance signal and the second virtual disturbance signal is a cosine virtual disturbance signal, and the amplitude of the virtual disturbance is small, the two differential responses can characterize the change in the spatial projection cost function along the corresponding virtual disturbance direction near the current rotor estimated angle. The first differential response, after being processed with the first virtual disturbance signal, forms a demodulation component related to the square of the sine term; the second differential response, after being processed with the second virtual disturbance signal, forms a demodulation component related to the square of the cosine term.

[0082] Specifically, since the first differential response represents the change in cost response after applying a first virtual disturbance based on the current rotor estimated angle, and the second differential response represents the change in cost response after applying a second virtual disturbance based on the current rotor estimated angle, when the amplitudes of the first and second virtual disturbance signals are small, the two differential responses can be regarded as local increments of the spatial projection cost function near the current rotor estimated angle. According to the Taylor expansion relation, the first dominant term of the cost response change of the spatial projection cost function near the current rotor estimated angle caused by the virtual disturbance is equal to the product of the gradient of the spatial projection cost function with respect to the current rotor estimated angle and the corresponding virtual disturbance amount. Therefore, the first dominant component of the first differential response corresponds to the product of the position error gradient and the first virtual disturbance signal, and the first dominant component of the second differential response corresponds to the product of the position error gradient and the second virtual disturbance signal. Furthermore, operating on the first differential response and the first virtual disturbance signal, and operating on the second differential response and the second virtual disturbance signal, is equivalent to synchronously demodulating the cost response changes caused by the two virtual disturbances. Thus, a term related to the square of the first virtual disturbance signal is formed in the first dominant component of the first demodulation result, and a term related to the square of the second virtual disturbance signal is formed in the first dominant component of the second demodulation result.

[0083] Add the first demodulation result and the second demodulation result together: ; According to Taylor's expansion theorem, the intrinsic mathematical equivalent of the above operation is: ; in, This is the magnetic flux observation vector. The magnitude of the magnetic flux observation vector. This is the rotor position error angle.

[0084] The orthogonality between the two virtual disturbance signals can be utilized, namely: =1; The original position error gradient is approximated as: ; In other words, the squared terms corresponding to the sinusoidal virtual disturbance in the first demodulation result and the squared terms corresponding to the cosine virtual disturbance in the second demodulation result are added together and combined into a constant 1 according to the trigonometric identity. This causes the high-frequency carrier component corresponding to the virtual disturbance frequency to be canceled out during the algebraic operation, while the remaining signal can characterize the gradient information of the rotor position error.

[0085] It should be noted that the carrier cancellation in this embodiment is achieved through the algebraic relationship between the two orthogonal virtual disturbance responses, rather than first obtaining the position error signal containing the high-frequency carrier and then filtering out the high-frequency components using a low-pass filter. Because: In algebraic operations, it directly equals 1, thus enabling the cancellation of the virtual high-frequency carrier within the current calculation cycle. This eliminates the need for further carrier extraction processing via a low-pass filter after demodulation of the position error gradient. Consequently, the low-pass filter module in traditional demodulation schemes can be omitted, avoiding the phase lag caused by the low-pass filter on the position error signal. This allows for the extraction of position error information with near-zero phase lag, improving the dynamic response capability of rotor position estimation to speed changes and load disturbances.

[0086] Through steps S401 and S402, the differential responses generated by the two orthogonal virtual disturbances are synchronously demodulated with the corresponding virtual disturbance signals. Then, by using the orthogonality of the sine and cosine signals for algebraic combination, the virtual carrier components can be canceled and the position error gradient can be obtained without the need for a low-pass filter. This provides error feedback information for subsequent normalization of the position error gradient and updating of the rotor estimated angle.

[0087] As one implementation method, such as Figure 5 As shown, step S105, which updates the rotor estimated angle based on the position error gradient, includes: Step S501. Determine the flux linkage amplitude parameter based on the first rotor flux linkage observation value and the second rotor flux linkage observation value.

[0088] Step S502. Normalize the position error gradient according to the magnetic flux amplitude parameter to obtain the normalized position error signal.

[0089] Step S503. Update the rotor estimated angle based on the normalized position error signal.

[0090] For step S501, the first rotor flux linkage observation value and the second rotor flux linkage observation value characterize the components of the flux linkage observation vector in the α-axis and β-axis directions, respectively. Therefore, the flux linkage amplitude parameter can be determined based on the sum of the squares of the two flux linkage observation components. For example, the flux linkage amplitude parameter can be determined according to the following formula: ; in, This represents the flux linkage amplitude parameter.

[0091] Therefore, the flux linkage observation information in both the α and β axes can be converted into scalar parameters related to the flux linkage vector magnitude, so that subsequent position error processing does not depend on the flux linkage component on a single coordinate axis.

[0092] Furthermore, when the motor is at zero speed or extremely low speed, the amplitude of the flux linkage observation vector can be limited to keep it within a preset range. For example, the amplitude of the flux linkage observation vector can be constrained to near a preset range corresponding to the permanent magnet flux linkage, preventing the amplitude from decaying to zero due to factors such as sampling DC bias and integral drift. This ensures that the flux linkage amplitude parameter used as the denominator in subsequent normalization processing remains non-zero, providing a stable amplitude basis for position error normalization under zero-speed conditions. This processing works in conjunction with the flux linkage amplitude limiting in the aforementioned hybrid flux linkage observation process.

[0093] For step S502, the position error gradient obtained by the aforementioned algebraic demodulation can be scaled using the flux linkage amplitude parameter determined in step S501, so as to reduce the influence of flux linkage amplitude change on the magnitude of the position error signal.

[0094] In one specific implementation, the normalized position error signal can be expressed as: ; in, This represents the normalized position error signal; Indicates the actual electrical angle of the rotor; Indicates the estimated electrical angle of the rotor; This indicates the position angle deviation between the actual electrical angle of the rotor and the estimated electrical angle of the rotor.

[0095] As can be seen from the above relationship, after normalization, the normalized position error signal is mainly related to the preset virtual disturbance amplitude and the position deviation between the actual electrical angle of the rotor and the estimated electrical angle of the rotor, which can reduce the influence of the change in the flux linkage observation amplitude on the position error feedback gain.

[0096] Furthermore, because the aforementioned flux linkage observation process employs a limit loop clamping mechanism, the flux linkage observation vector can maintain a preset non-zero amplitude even at zero or extremely low speeds. Therefore: ; This avoids the division-to-zero singularity caused by the denominator being zero during the normalization process, and improves the stability of position error calculation under zero velocity conditions.

[0097] In addition, by: ; It can be seen that the flux linkage amplitude correlation term in the normalized position error signal is eliminated, so that the equivalent gain of the position tracking closed loop is mainly determined by the preset virtual disturbance amplitude. The position tracking closed-loop gain is fixed and no longer changes significantly with variations in flux amplitude or rotational speed. In other words, the position tracking closed-loop gain can be decoupled from the motor operating speed, maintaining an effective error feedback gain even at zero speed. This avoids the problem in traditional position estimation methods based on back EMF where the feedback gain approaches zero as the rotational speed approaches zero.

[0098] Therefore, at the moment of motor startup or when the motor is at zero speed, the normalized position error signal can still reflect the deviation between the actual rotor position and the estimated rotor position, so that the subsequent position tracking stage can correct the estimated rotor angle based on the position deviation, thus providing conditions for establishing an effective position estimation closed loop at zero speed.

[0099] For step S503, the normalized position error signal obtained in step S502 can be used as the error input of the position tracking closed loop, and the current rotor estimated angle can be adjusted in a closed loop according to the normalized position error signal.

[0100] Specifically, when there is a deviation between the actual electrical angle of the rotor and the estimated electrical angle of the rotor, the normalized position error signal is: ; It can characterize the positional deviation between the two. The position tracking stage can determine the adjustment direction of the rotor estimated angle based on this normalized position error signal, and continuously update the rotor estimated angle so that the rotor estimated electrical angle gradually approaches the rotor actual electrical angle.

[0101] Through steps S501 to S503, on the one hand, the original position error gradient is normalized by the sum of squares of the flux linkage observation components, and combined with the limit loop clamping mechanism in the flux linkage observation process, the normalized denominator remains non-zero in the zero-speed state, thereby improving the system's robustness to division-zero singularities. On the other hand, the normalization process makes the equivalent gain of the position tracking closed loop mainly determined by the preset virtual disturbance amplitude, thereby reducing the dependence of the position feedback gain on the flux linkage amplitude and motor speed, so that the motor can still form effective position error feedback in the zero-speed start-up stage, providing a basis for realizing rotor position closed-loop estimation in the zero-speed state.

[0102] As one implementation method, such as Figure 6 As shown, step S503, which updates the rotor estimated angle based on the normalized position error signal, includes: Step S601. Input the normalized position error signal into the third-order extended state observer, and use the third-order extended state observer to observe the state of the normalized position error signal to obtain the rotor angle state, rotor speed state and rotor acceleration state.

[0103] Step S602. Update the estimated rotor angle based on the rotor angle status.

[0104] Among them, the normalized position error signal obtained by the aforementioned normalization process can be used. A third-order extended state observer (Type-3 ESO) is input to perform closed-loop correction of the rotor motion state using the normalized position error signal. The third-order extended state observer can set three sequentially related state variables to characterize the rotor angle, rotor speed, and rotor acceleration, respectively. This allows the position estimation process to not only correct the rotor angle based on the current position error, but also to track the rotor motion trend by combining changes in rotor speed and rotor acceleration.

[0105] In one specific implementation, the state update equation of the third-order extended state observer can be expressed as:

[0106] in, This represents the rotor angle state, i.e., the rotor estimated electrical angle output by the third-order extended state observer; This indicates the rotor speed state, i.e., the estimated rotor electrical angular velocity; This indicates the rotor acceleration state, i.e., the estimated rotor electrical angular acceleration; This represents the normalized position error signal; This represents the gain of the first observer acting on the rotor angle state update process. This represents the gain of the second observer acting on the rotor speed state update process; The gain of the third observer is used to influence the rotor acceleration state update process; t represents time.

[0107] Specifically, the first state update equation updates the rotor angle state using the current rotor speed state and the normalized position error signal; the second state update equation updates the rotor speed state using the current rotor acceleration state and the normalized position error signal; and the third state update equation updates the rotor acceleration state using the normalized position error signal. Thus, the three state variables form a hierarchically correlated state observation structure, enabling the normalized position error signal to simultaneously participate in the correction of rotor angle, rotor speed, and rotor acceleration.

[0108] Furthermore, the observer gain of the third-order extended state observer can be configured using a bandwidth separation approach. In one specific implementation, the main tracking pole of the third-order extended state observer can be configured as a preset observation bandwidth ωo, and the first observer gain and the second observer gain can be determined based on the preset observation bandwidth, for example: β1=3ωo; β2=3ωo; Where ωo represents the preset observation bandwidth of the third-order extended state observer. Indicates the gain of the first observer. This indicates the gain of the second observer.

[0109] Third observer gain The bandwidth is configured independently and set to a smaller gain value. This bandwidth separation configuration allows for relatively fast tracking responses for both rotor angle and rotor speed states, while simultaneously enabling observation of rotor acceleration states with relatively independent gain, thus balancing the speed of position tracking with the stability of dynamic state observation.

[0110] Furthermore, the rotor speed and rotor acceleration states in the third-order extended state observer can provide information related to the rotor motion trend. When the motor is under dynamic conditions such as rapid acceleration, rapid deceleration, or sudden load changes, the rotor acceleration state can characterize the rapid change in rotor speed and act on the rotor speed and rotor angle states through the aforementioned state update relationship. Therefore, the higher-order motion states of the third-order extended state observer can be used to track the rotor position change trend, thereby reducing the tracking deviation between the estimated rotor angle and the actual rotor angle during rapid acceleration and deceleration of the motor.

[0111] Furthermore, the third-order extended state observer has two zeros located in the left half-plane, which can be used to form kinematic feedforward lead compensation to further suppress phase tracking errors generated during rapid motor acceleration. By configuring the observation bandwidth corresponding to the main tracking pole and the gain of the third observer corresponding to the rotor acceleration state relatively independently, the position tracking response and acceleration observation response can be adjusted separately according to the actual dynamic characteristics of the motor.

[0112] Within each control cycle, the rotor angle state, rotor speed state, and rotor acceleration state can be iteratively updated according to the aforementioned state update equation, and the updated rotor angle state is used as the final estimated rotor angle obtained in the current control cycle. This estimated rotor angle can be fed back to the aforementioned spatial projection cost function and the biorthogonal virtual disturbance loop to continue cost response calculation, position error gradient extraction, and position closed-loop correction in the next control cycle; on the other hand, it can serve as the rotor position information required for the field-oriented control of the permanent magnet synchronous motor.

[0113] Therefore, by inputting the normalized position error signal into the third-order extended state observer, state information at three levels—rotor angle, rotor speed, and rotor acceleration—can be obtained simultaneously. The rotor speed and rotor acceleration states can be used to improve the tracking ability of rotor motion trends. Combined with the bandwidth-separated observer parameter configuration, the dynamic motion state can be observed while maintaining the ability to track rotor position quickly, thereby reducing the position tracking error caused by rapid acceleration and deceleration of the motor and sudden load changes.

[0114] As one implementation method, such as Figure 7As shown, controlling the motor operation in step S105 includes: Step S701. Perform Park transformation and inverse Park transformation based on the updated rotor estimated angle to perform field-oriented control of the motor.

[0115] Step S702. Generate a pulse width modulation signal based on the magnetic field orientation control result to control the inverter to drive the motor.

[0116] After obtaining the updated rotor estimated angle through the third-order extended state observer, the rotor estimated angle can be used as the rotor position information required by field-oriented control (FOC) and input into the Park transformation module and the inverse Park transformation module to establish the coordinate transformation relationship between the two-phase stationary coordinate system on the stator side of the motor and the two-phase rotating coordinate system that rotates with the rotor.

[0117] Specifically, the Park transformation can be performed on the motor current in the two-phase stationary coordinate system based on the updated rotor estimated angle, converting the current components in the two-phase stationary coordinate system into d-axis and q-axis current components in the two-phase rotating coordinate system. The d-axis current component represents the current component related to the rotor flux direction, while the q-axis current component represents the current component related to the electromagnetic torque. By transforming the motor current to a coordinate system that rotates synchronously with the rotor, closed-loop regulation can be performed on the corresponding current components.

[0118] Furthermore, current closed-loop calculations can be performed based on the deviations between the d-axis and q-axis current components and their corresponding current setpoints, thereby obtaining the d-axis and q-axis reference voltages. This allows for the control of the motor's flux linkage and torque-related components in a two-phase rotating coordinate system, achieving field-oriented control.

[0119] After obtaining the d-axis and q-axis reference voltages, an inverse Park transformation can be performed using the updated rotor estimated angles to convert the d-axis and q-axis reference voltages in the two-phase rotating coordinate system into reference voltages in the two-phase stationary coordinate system. For example, the α-axis and β-axis reference voltages can be obtained for use in subsequent pulse width modulation stages.

[0120] The rotor position information used in both the Park transform and the inverse Park transform can be the updated rotor estimated angle output by the aforementioned third-order extended state observer. Therefore, the coordinate relationship between stator current, voltage, and rotor flux direction can be established without obtaining the rotor position through a mechanical position sensor.

[0121] Furthermore, pulse width modulation (PWM) calculations can be performed based on the reference voltage in the two-phase stationary coordinate system obtained by the inverse Park transformation to obtain a PWM signal for controlling the inverter. This PWM signal can be used to determine the on and off states of each power switching device in the inverter, enabling the inverter to convert DC-side electrical energy into AC electrical energy required to drive the motor according to the corresponding switching combinations.

[0122] The microprocessor can output the generated pulse width modulation signal to the inverter, and control the power switching devices in the inverter to turn on and off according to the corresponding duty cycle, so that the inverter outputs a drive voltage to the motor corresponding to the field orientation control result, thereby adjusting the stator current of the motor and controlling the motor operation.

[0123] Furthermore, after the inverter drives the motor to start, the stator current and reference voltage of the motor can be reacquired in the next control cycle, and the flux linkage observation value is updated based on the newly acquired stator current and reference voltage. Then, spatial projection cost calculation, biorthogonal virtual disturbance, differential response calculation, algebraic demodulation, position error normalization, and third-order extended state observation are performed sequentially to obtain the updated rotor estimated angle again. The updated rotor estimated angle continues to participate in the Park transform and inverse Park transform in the next control cycle, thus forming a closed-loop control process from motor operating parameter acquisition, flux linkage observation, position error extraction, rotor angle update, FOC current closed loop, and PWM drive.

[0124] In this way, the rotor estimated angle obtained by the aforementioned sensorless algorithm can be directly applied to field-oriented control, realizing the motor current closed loop and inverter drive without the need for a mechanical position sensor to provide the rotor position, and enabling the rotor position estimation process and the motor operation control process to continuously iterate in a closed loop.

[0125] like Figure 8 The figure shows the amplitude characteristic curves of the closed-loop third-order extended state observer provided in this embodiment under different observation bandwidths. In the lower frequency range, the amplitude response corresponding to different observation bandwidths remains close to 0 dB, indicating that the third-order extended state observer can effectively track position error information in the low frequency range.

[0126] As the input signal frequency increases, the amplitude response gradually decreases, indicating that the third-order extended state observer has an attenuation effect on higher frequency signals. Therefore, while completing rotor angle state tracking, the third-order extended state observer can also suppress higher frequency components in the normalized position error signal, thus playing a certain filtering role.

[0127] Furthermore, the curves corresponding to different observation bandwidths show that the curves corresponding to higher observation bandwidths maintain a higher amplitude response over a higher frequency range. This indicates that increasing the observation bandwidth of the third-order extended state observer can improve its ability to track rapidly changing signals and increase the response speed of the position estimation closed loop.

[0128] like Figure 9 The figure shows the frequency response curves of the closed-loop third-order extended state observer provided in this embodiment under different observation bandwidths. In the low-frequency range, the phase corresponding to different observation bandwidths is close to 0°, indicating that in this frequency range, the output of the third-order extended state observer has a small phase deviation relative to the input position error information.

[0129] As the frequency increases, the phase of each curve gradually decreases and approaches approximately -90°, indicating that as the rate of change of the input signal increases, the output of the third-order extended state observer gradually produces phase lag.

[0130] Meanwhile, it can be observed that as the observation bandwidth increases, the frequency at which the phase begins to decrease significantly shifts towards higher frequencies. In other words, at the same input signal frequency, a larger observation bandwidth can typically maintain a smaller phase lag over a wider frequency range, thus improving the rotor position estimation's ability to track fast dynamic processes.

[0131] The motor control method provided in this embodiment combines flux linkage observation, spatial projection cost calculation, biorthogonal virtual disturbance, algebraic demodulation, error normalization, and third-order extended state observation to achieve rotor position estimation and closed-loop control of a permanent magnet synchronous motor at zero speed. This allows the motor to directly enter closed-loop operation from zero speed without open-loop drive and adapts to starting requirements under different load conditions. It can achieve smooth starting even under load or full load conditions, reducing the possibility of reversal or starting failure during startup. Simultaneously, through spatial geometric perturbation mapping and bounded constraints on internal error state variables, it can suppress error accumulation and integral saturation caused by large initial rotor position estimation errors, and limit the inverter startup output current within a preset safety range, thereby reducing startup overcurrent and the resulting damage to power devices. This reduces risks and improves the starting safety and reliability of the motor and inverter. Furthermore, the first and second virtual disturbance signals only act on the rotor estimated angle within the digital control domain, eliminating the need to actually inject physical high-frequency voltage into the motor windings. This reduces high-frequency electromagnetic noise and additional losses caused by physical high-frequency signal injection. Further, by algebraically combining the differential responses corresponding to the two orthogonal virtual disturbances, the orthogonality between sine and cosine signals is used to directly cancel the virtual carrier component. This eliminates the need for a low-pass filter to filter the position error signal, thus avoiding the group delay and phase lag introduced by the low-pass filter. This improves the dynamic response performance of position error information extraction and rotor position closed-loop tracking, thereby enhancing the sensorless control performance of the permanent magnet synchronous motor during zero-speed start-up, load start-up, and dynamic operation.

[0132] like Figure 10 As shown, Figure 10 The experimental results of direct zero-speed closed-loop start using the motor control method provided in this embodiment are shown under 50% rated load conditions and with inductance parameter mismatch. Figure 10 The data, from top to bottom, shows the changes in the motor's phase current, actual speed, rotor angle, and rotor position error over time, with the horizontal axis representing time.

[0133] Specifically, by Figure 10 The phase current changes show that during the motor's start-up from zero speed and gradual acceleration, the phase current establishes itself during the startup process and maintains a periodic change after the motor enters a stable operating state. No continuously increasing current was observed during the startup process. Figure 10 The actual speed change shows that the motor can directly enter the acceleration process from zero speed, and the speed gradually rises to near the target speed. After the start-up is completed, it stabilizes at near the target speed. This indicates that even under the conditions of 50% rated load and mismatch in inductance parameters, the motor control method provided in this embodiment can still achieve closed-loop start-up from zero speed.

[0134] Furthermore, by Figure 10 The rotor angle changes show that in the initial startup phase, there is a certain positional deviation between the estimated rotor angle and the actual rotor angle. As the position estimation closed loop is gradually established, the estimated rotor angle gradually tracks the actual rotor angle, and after the motor enters the stable operation phase, it becomes basically synchronized with the actual rotor angle. Correspondingly, Figure 10 The rotor position error exhibits significant variation during the initial startup phase, gradually decreasing as the estimated rotor angle is continuously corrected, eventually converging to near zero and fluctuating within a small range. This demonstrates that the motor control method provided in this embodiment can continuously correct the estimated rotor angle even with initial position deviations, gradually bringing the estimated rotor position closer to the actual rotor position and establishing a stable position estimation closed loop.

[0135] Furthermore, to verify the adaptability of the motor control method provided in this embodiment to motor parameter mismatch, the preset motor parameters in the control algorithm can be made to deviate to different degrees from the actual motor parameters. Experimental results show that when there is a ±50% parameter mismatch in the preset stator inductance, ±50% parameter mismatch in the permanent magnet flux linkage, and +30% to 50% parameter mismatch in the stator resistance, the motor control method provided in this embodiment can still complete zero-speed closed-loop start-up. This demonstrates that the motor control method is adaptable to motor parameter mismatches such as stator inductance, permanent magnet flux linkage, and stator resistance within a certain range, and can still maintain rotor position estimation and zero-speed closed-loop start-up process even when there are deviations between the motor parameters and the preset parameters in the control algorithm.

[0136] therefore, Figure 10 The experimental results show that the motor control method provided in this embodiment can still directly establish a rotor position estimation closed loop and a motor drive closed loop from zero speed under conditions of 50% rated load and certain mismatch in motor parameters, so that the motor speed gradually reaches the target speed and the rotor position error gradually converges. This improves the adaptability and robustness of the sensorless control method to load start-up and motor parameter mismatch conditions.

[0137] like Figure 11 As shown, Figure 11The experiment shows the results of direct zero-speed closed-loop start-up of a permanent magnet synchronous motor under 100% full load conditions using the motor control method provided in this embodiment. The DC bus voltage used in the experiment can be set to 24V, the PWM switching frequency to 10kHz, the virtual disturbance frequencies corresponding to the first and second virtual disturbance signals to 1000Hz, the virtual disturbance amplitude to 0.1rad, and the target speed to 100r / min, to verify the performance of the motor control method provided in this embodiment in directly establishing closed-loop control from zero speed under full load conditions.

[0138] Specifically, Figure 11 The diagram, from top to bottom, shows the changes in phase current, actual speed, rotor angle, and rotor position error of the motor during startup over time, with the horizontal axis representing time. During the experiment, with the motor initially at zero speed and under 100% full load, the aforementioned flux linkage observation, spatial projection cost calculation, biorthogonal virtual disturbance, algebraic demodulation, position error normalization, and third-order extended state observation processes were directly initiated. Magnetic field orientation control was then performed using the updated rotor estimated angle to allow the motor to directly enter the closed-loop startup process from zero speed.

[0139] Depend on Figure 11 The phase current changes show that during full-load zero-speed startup, as the motor gradually builds up electromagnetic torque, the amplitude of the phase current increases accordingly to generate the starting torque required to overcome the load torque and static friction. After the motor completes startup and enters a stable operating phase, the phase current exhibits a stable periodic change. Throughout the entire startup process, the phase current remains within the corresponding control range, and no abnormally increasing current occurs, indicating that the corresponding current closed-loop control can be maintained during full-load startup.

[0140] Depend on Figure 11 The actual speed changes show that the motor can directly enter the startup process under 100% full load and initial speed of zero. In the initial startup phase, the motor speed exhibits some transient changes, then continuously increases and gradually reaches the target speed of around 100 r / min. After reaching the target speed, the actual speed fluctuates within a certain range around the target speed and remains stable. This indicates that the motor control method provided in this embodiment does not require first accelerating the motor to the preset speed via open-loop drive before switching to closed-loop control; instead, it can directly establish a closed loop and complete the startup under full load conditions from zero speed.

[0141] Furthermore, by Figure 11The rotor angle changes show that in the initial startup phase, there is a certain positional deviation between the estimated rotor angle and the actual rotor angle; as the position estimation closed loop is gradually established, the estimated rotor angle rapidly approaches the actual rotor angle and maintains this alignment during subsequent operation. Correspondingly, Figure 11 The rotor position error exhibits significant variation during the initial startup phase. As the position error gradient is continuously used to correct the estimated rotor angle, the rotor position error decreases rapidly and converges to near zero after approximately 2 seconds, subsequently fluctuating within a small range. This demonstrates that even under 100% full-load startup conditions, the position estimation method provided in this embodiment can still establish effective position feedback, gradually locking the estimated rotor angle to the actual rotor position.

[0142] Furthermore, direct zero-speed closed-loop start-up experiments were conducted under 0 load, 50% rated load, and 100% full load conditions. Experimental results show that, under different load conditions, the motor control method provided in this embodiment can directly establish position estimation and motor drive closed loops from zero speed. Specifically, under 100% full load conditions, the phase current can generate the starting torque required to overcome the load and static friction, the rotor estimation angle can gradually lock the actual rotor position during startup, and the motor speed gradually reaches the target speed without continuous reverse rotation or startup divergence.

[0143] therefore, Figure 11 The experimental results show that the motor control method provided in this embodiment can achieve direct zero-speed closed-loop start-up under 100% full load conditions, and gradually converge the rotor position error during the start-up process, bringing the actual speed close to the target speed. Therefore, it can improve the position estimation capability and start-up adaptability of sensorless control of permanent magnet synchronous motors under high-load start-up conditions, providing a foundation for direct zero-speed closed-loop start-up of motors under different load conditions.

[0144] like Figure 12 As shown, Figure 12 This demonstrates the permanent magnet synchronous motor rapidly reversing from +100 r / min to... The results of the 100 r / min speed reversal experiment were used to verify the position estimation and closed-loop tracking performance of the motor control method provided in this embodiment when the motor speed crosses the zero-speed region and the rotation direction changes. Figure 8 The data, from top to bottom, shows the changes in phase current, actual speed, rotor angle, and rotor position error over time during the speed reversal process, with the horizontal axis representing time.

[0145] Specifically, in the speed reversal experiment, the motor can first be controlled to run stably at a speed of +100 r / min. Then, a reverse speed command is applied to rapidly decelerate the motor from the forward rotation state. After the actual speed drops to zero, it continues to accelerate in the opposite direction until the actual speed reaches zero. Around 100 r / min. (By) Figure 12 The actual speed change shows that, under the reverse command, the motor's actual speed drops rapidly from +100 r / min and crosses the zero-speed region, then accelerates in the opposite direction to... The motor operates at around 100 r / min and remains stable after reversing, indicating that the motor control method provided in this embodiment can achieve continuous switching between the forward and reverse running states of the motor.

[0146] Furthermore, by Figure 12 The phase current changes show that during the stable forward and reverse operation phases of the motor, the phase current maintains a periodic change. During the dynamic processes of motor deceleration, crossing zero speed, and reverse acceleration, the phase current undergoes corresponding transient changes to generate the electromagnetic torque required for motor deceleration and reverse acceleration, and then re-enters a relatively stable periodic change state after the reversal is completed. Therefore, the motor can maintain a corresponding torque output during rapid speed direction switching.

[0147] Depend on Figure 12 The changes in rotor angle and rotor position error show that when the motor is running stably at +100 r / min, the estimated rotor angle can track the actual rotor angle, and the rotor position error remains near zero. When the motor begins to reverse rapidly and crosses the zero-speed region, the dynamic change of the rotor angle increases, and the rotor position error produces transient fluctuations accordingly. After the motor crosses the zero-speed region and enters the reverse operation state, the rotor position error can decrease again and stabilize within a small range, and the estimated rotor angle resumes tracking the actual rotor angle.

[0148] Specifically, during the transition of the motor from forward to reverse rotation, the actual rotational speed inevitably crosses the zero-speed region. Within this region, position observation methods based on back electromotive force (EMF) are prone to decreased observability of rotor position information because the back EMF decreases with decreasing rotational speed. According to... Figure 12 The experimental results show that the motor control method provided in this embodiment can still maintain the position estimation closed loop during the process of crossing the zero speed region. Although the rotor position error fluctuates briefly at the moment of speed reversal, it can converge again after crossing the zero speed region, and there is no continuous 180° erroneous position convergence.

[0149] In this embodiment, flux linkage observation information in a two-phase stationary coordinate system is obtained through flux linkage observation. Based on a spatial projection cost function, mutually orthogonal virtual disturbances are applied in the digital control domain to obtain a cost response related to rotor position deviation. Furthermore, the position error gradient is obtained through differential processing and algebraic demodulation, and normalized using the flux linkage amplitude parameter. Therefore, when the actual motor speed is close to zero or crosses the zero-speed region, the position error feedback does not directly depend on the actual back EMF amplitude. The obtained position error gradient can be used to further correct the rotor estimated angle, thereby reducing the possibility of position estimation failure near zero speed and convergence of erroneous angles.

[0150] Furthermore, the normalized position error signal can be further input into a third-order extended state observer, which jointly observes the rotor angle state, rotor speed state, and rotor acceleration state. This is done when the motor rapidly reverses from +100 r / min to... During the process of 100 r / min, the rotor speed and rotor acceleration change significantly. By jointly observing the above motion states, the rotor estimated angle can be adapted to the continuous dynamic process of rapid deceleration, zero-speed crossing and reverse acceleration, and a stable position tracking state can be re-established after the reversal is completed.

[0151] therefore, Figure 12 The experimental results shown demonstrate that the motor control method provided in this embodiment can rapidly reverse the motor from +100 r / min to... The rotor position is continuously estimated during the 100 r / min process and can cross the zero-speed region; the rotor position error generated during the speed reversal process can be re-converged after the reversal, thereby reducing the 180° erroneous convergence problem that may occur in the traditional zero-speed position observation process, and improving the stability and continuity of sensorless control of permanent magnet synchronous motor under forward and reverse switching, zero-speed crossing and fast dynamic operation conditions.

[0152] like Figure 13 As shown, Figure 13 The experimental results of stable operation of a permanent magnet synchronous motor at extremely low speeds using the motor control method provided in this embodiment are shown. This verifies the rotor position estimation and closed-loop control performance of the motor control method provided in this embodiment under low speed, low signal-to-noise ratio, and mechanical disturbance conditions. Figure 13 The data, from top to bottom, shows the changes in the motor's phase current, actual speed, and rotor angle over time, with the horizontal axis representing time.

[0153] Specifically, in the ultra-low speed steady-state experiment, the target speed of the permanent magnet synchronous motor can be set to 10 r / min, and closed-loop operation can be continuously performed at this speed. Figure 13The phase current changes show that when the motor is running at an extremely low speed of 10 r / min, the phase current maintains a continuous periodic change and there is no obvious current divergence, indicating that the corresponding current closed-loop control can still be maintained under extremely low speed conditions.

[0154] Depend on Figure 13 The actual speed changes show that when the target speed is set to 10 r / min, the actual motor speed can be maintained near the target speed. Due to mechanical disturbances such as cogging torque that may occur during extremely low-speed operation, the actual speed fluctuates to a certain extent near the target speed. However, no continuously increasing speed deviation or closed-loop instability occurred during the experiment, indicating that the motor control method provided in this embodiment can maintain continuous motor operation under extremely low-speed conditions with mechanical disturbances.

[0155] Furthermore, by Figure 13 The rotor angle change shows that during motor operation at 10 r / min, the rotor angle changes continuously with time and periodically according to the actual rotor rotation process, indicating that rotor position information for motor closed-loop control can still be continuously obtained even at extremely low speeds. Although the motor speed is low at this time, and the signal amplitude that can be used for position observation based on back EMF is correspondingly reduced, this embodiment obtains flux linkage observation information through flux linkage observation and extracts the position error gradient in the digital control domain using the spatial projection cost function and mutually orthogonal virtual disturbance signals. Therefore, it can reduce the dependence of the position error information extraction process on the actual back EMF amplitude.

[0156] Furthermore, by normalizing the position error gradient using the flux linkage amplitude parameter and inputting the normalized position error signal into a third-order extended state observer, the rotor angle, rotor speed, and rotor acceleration states can be continuously updated. This allows the estimated rotor angle to track the rotor's motion even at extremely low speeds. Therefore, even under periodic mechanical disturbances caused by factors such as cogging torque, the position estimation closed loop can continue to operate, thus maintaining the motor in extremely low-speed closed-loop operation.

[0157] Furthermore, since the first and second virtual disturbance signals only virtually disturb the rotor's estimated angle within the digital control domain without applying physical high-frequency voltages to the motor windings, rotor position information can be obtained without relying on physical high-frequency signal injection during extremely low-speed continuous operation, reducing high-frequency electromagnetic noise and additional losses caused by physical high-frequency signal injection. Simultaneously, the position error gradient is directly obtained through algebraic demodulation of the differential responses corresponding to the two orthogonal virtual disturbances, eliminating the need to extract position error information through a low-pass filter. This reduces the phase lag introduced by the low-pass filter, improving the response capability of the position estimation closed loop to mechanical disturbances at extremely low speeds.

[0158] therefore, Figure 13 The experimental results show that the motor control method provided in this embodiment can maintain continuous closed-loop operation at an extremely low speed of about 10 r / min. Even with mechanical disturbances caused by cogging torque and weak low-speed position signals, the actual speed can still be maintained near the target speed, and the rotor angle can be continuously tracked. This indicates that the motor control method has good adaptability and anti-disturbance capability for extremely low-speed operation.

[0159] Example 2 This second embodiment provides a motor control device, such as... Figure 14 As shown, the motor control device includes: The rotor flux linkage observation acquisition module 101 is used to acquire the stator current, reference voltage and rotor estimated angle of the motor in a two-phase stationary coordinate system. The rotor flux linkage observation value of the motor is acquired based on the stator current and reference voltage. The rotor flux linkage observation value includes the first rotor flux linkage observation value and the second rotor flux linkage observation value. The cost response acquisition module 102 is used to construct a spatial projection cost function, determine the first cost response corresponding to the rotor estimated angle when no virtual disturbance is applied based on the first rotor flux observation value, the second rotor flux observation value and the rotor estimated angle; generate a first virtual disturbance signal and a second virtual disturbance signal that are mutually orthogonal, determine the second cost response based on the spatial projection cost function and the first virtual disturbance signal, and determine the third cost response based on the spatial projection cost function and the second virtual disturbance signal. The differential response acquisition module 103 is used to determine the first differential response based on the second cost response and the first cost response, and to determine the second differential response based on the third cost response and the first cost response; The position error gradient acquisition module 104 is used to perform algebraic demodulation based on the first differential response, the second differential response, the first virtual disturbance signal, and the second virtual disturbance signal to obtain the position error gradient. The motor control module 105 is used to update the rotor estimated angle according to the position error gradient in order to control the motor operation.

[0160] The rotor flux linkage observation acquisition module 101 is used to collect the stator current and reference voltage of the motor in a two-phase stationary coordinate system, and combine them with the corresponding motor parameters to perform flux linkage observation, so as to obtain the first rotor flux linkage observation value and the second rotor flux linkage observation value, and provide flux linkage information for the subsequent extraction of rotor position error information.

[0161] The cost response acquisition module 102 is used to construct a spatial projection cost function based on the first rotor flux linkage observation value, the second rotor flux linkage observation value, and the current estimated rotor angle, and to calculate the first cost response corresponding to the absence of virtual disturbance. Simultaneously, the cost response acquisition module generates mutually orthogonal first and second virtual disturbance signals in the digital control domain, and determines the second and third cost responses based on the two virtual disturbances respectively, thereby obtaining spatial projection results under different virtual disturbance states.

[0162] The differential response acquisition module 103 is used to perform differential processing on the second and third cost responses, respectively, using the first cost response as a reference in the unperturbed state, to obtain the first differential response and the second differential response. Through the above differential processing, the changes in cost response caused by the first virtual perturbation signal and the second virtual perturbation signal can be extracted.

[0163] The position error gradient acquisition module 104 is used to perform algebraic demodulation on the first differential response and the second differential response with the corresponding first virtual disturbance signal and the second virtual disturbance signal, respectively, and to cancel the corresponding carrier components by utilizing the orthogonality between the two virtual disturbance signals, so as to obtain the position error gradient characterizing the rotor position deviation. Therefore, the position error gradient can be obtained without using a low-pass filter to extract position error information.

[0164] The motor control module 105 is used to update the current estimated rotor angle based on the position error gradient, and to perform closed-loop control of the motor based on the updated estimated rotor angle. In some embodiments, the motor control module can also normalize the position error gradient and input the normalized position error signal into a third-order extended state observer to obtain the updated estimated rotor angle; further, it performs Park transform and inverse Park transform based on the updated estimated rotor angle to complete field-oriented control and generate a pulse width modulation signal to control the inverter to drive the motor.

[0165] It should be noted that the motor control device provided in Embodiment 2 is based on the same technical concept as the motor control method provided in Embodiment 1. The specific processing procedures and corresponding technical effects of each module can be referred to the relevant descriptions in Embodiment 1, and will not be repeated here.

[0166] This application also provides a readable storage medium storing a computer program, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0167] This application provides a computer program product that stores a computer program, which includes program instructions. When executed by a computer, the program instructions cause the computer to perform the steps described in the various method embodiments above.

[0168] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. A computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0169] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0170] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0171] In the embodiments provided in this application, it should be understood that the disclosed apparatus / device and method can be implemented in other ways. For example, the apparatus / device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0172] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0173] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A motor control method, characterized in that, The motor control method includes: The stator current, reference voltage, and estimated rotor angle of the motor are obtained in a two-phase stationary coordinate system. The rotor flux linkage observation value of the motor is obtained based on the stator current and the reference voltage. The rotor flux linkage observation value includes a first rotor flux linkage observation value and a second rotor flux linkage observation value. A spatial projection cost function is constructed, and based on the first rotor flux observation value, the second rotor flux observation value, and the estimated rotor angle, the first cost response corresponding to the estimated rotor angle when no virtual disturbance is applied is determined through the spatial projection cost function. A first virtual disturbance signal and a second virtual disturbance signal that are orthogonal to each other are generated, and a second cost response is determined by the spatial projection cost function and the first virtual disturbance signal, and a third cost response is determined by the spatial projection cost function and the second virtual disturbance signal; A first differential response is determined based on the second cost response and the first cost response, and a second differential response is determined based on the third cost response and the first cost response; Algebraic demodulation is performed based on the first differential response, the second differential response, the first virtual disturbance signal, and the second virtual disturbance signal to obtain the position error gradient, and the rotor estimated angle is updated based on the position error gradient to control the operation of the motor.

2. The motor control method according to claim 1, characterized in that, The step of obtaining the observed rotor flux linkage of the motor based on the stator current and the reference voltage includes: Dynamic back electromotive force is observed based on the stator current, the reference voltage, and the motor parameters of the motor to obtain the observed dynamic back electromotive force value. The rotor flux linkage observation value is obtained by performing hybrid flux linkage observation based on the dynamic back electromotive force observation value and the steady-state current model reference value.

3. The motor control method according to claim 1, characterized in that, The spatial projection cost function is: ; in, This is the first rotor flux linkage observation value. To estimate the angle of the rotor, To estimate the α-axis direction component corresponding to the angle, This is the second rotor flux linkage observation value. To estimate the β-axis direction component corresponding to the angle.

4. The motor control method according to claim 1, characterized in that, Determining a second cost response using the spatial projection cost function and the first virtual perturbation signal, and determining a third cost response using the spatial projection cost function and the second virtual perturbation signal, includes: The rotor estimated angle is perturbed according to the first virtual disturbance signal to obtain a first disturbance angle, and the rotor estimated angle is perturbed according to the second virtual disturbance signal to obtain a second disturbance angle. The second cost response is determined based on the spatial projection cost function and the first disturbance angle, and the third cost response is determined based on the spatial projection cost function and the second disturbance angle.

5. The motor control method according to claim 1, characterized in that, The step of performing algebraic demodulation based on the first differential response, the second differential response, the first virtual disturbance signal, and the second virtual disturbance signal to obtain the position error gradient includes: A first demodulation result is determined based on the first differential response and the first virtual disturbance signal, and a second demodulation result is determined based on the second differential response and the second virtual disturbance signal. The first demodulation result and the second demodulation result are added together, and the corresponding carrier components are canceled out according to the orthogonality between the first virtual disturbance signal and the second virtual disturbance signal to obtain the position error gradient.

6. The motor control method according to claim 1, characterized in that, The step of updating the rotor estimated angle based on the position error gradient includes: The flux linkage amplitude parameter is determined based on the first rotor flux linkage observation value and the second rotor flux linkage observation value; The position error gradient is normalized according to the magnetic flux amplitude parameter to obtain a normalized position error signal; The rotor estimated angle is updated based on the normalized position error signal.

7. The motor control method according to claim 6, characterized in that, Updating the rotor estimated angle based on the normalized position error signal includes: The normalized position error signal is input into a third-order extended state observer, and the state of the normalized position error signal is observed through the third-order extended state observer to obtain the rotor angle state, rotor speed state and rotor acceleration state. The estimated rotor angle is updated based on the rotor angle state.

8. The motor control method according to claim 7, characterized in that, The control of the motor operation includes: Based on the updated rotor estimated angle, Park transformation and inverse Park transformation are performed to perform field-oriented control of the motor; A pulse width modulation signal is generated based on the magnetic field orientation control result to control the inverter to drive the motor.

9. A motor control device, characterized in that, The motor control device includes: The rotor flux linkage observation module is used to acquire the stator current, reference voltage and rotor estimated angle of the motor in a two-phase stationary coordinate system, and to acquire the rotor flux linkage observation value of the motor based on the stator current and the reference voltage. The rotor flux linkage observation value includes a first rotor flux linkage observation value and a second rotor flux linkage observation value. The cost response acquisition module is used to construct a spatial projection cost function, determine a first cost response corresponding to the estimated rotor angle when no virtual disturbance is applied based on the first rotor flux linkage observation value, the second rotor flux linkage observation value, and the estimated rotor angle; generate a first virtual disturbance signal and a second virtual disturbance signal that are mutually orthogonal, determine a second cost response based on the spatial projection cost function and the first virtual disturbance signal, and determine a third cost response based on the spatial projection cost function and the second virtual disturbance signal. The differential response acquisition module is used to determine a first differential response based on the second cost response and the first cost response, and to determine a second differential response based on the third cost response and the first cost response; The position error gradient acquisition module is used to perform algebraic demodulation based on the first differential response, the second differential response, the first virtual disturbance signal, and the second virtual disturbance signal to obtain the position error gradient. The motor control module is used to update the rotor estimated angle according to the position error gradient in order to control the operation of the motor.

10. A computer program product, characterized in that, The computer program product stores a computer program, the computer program including program instructions, which, when executed by a computer, cause the computer to perform the method as described in any one of claims 1 to 8.