Low-noise flux linkage adaptive feedback control system of high-speed motor

Abstract

The invention relates to the technical field of motor control, and discloses a low-noise flux linkage adaptive feedback control system of a high-speed motor. The system comprises an acquisition module used for acquiring current, voltage and vibration data of a high-speed motor at various operation rotating speeds to obtain operation parameters and noise characteristic data of the high-speed motor; the transformation processing module is used for performing frequency domain orthogonal transformation processing on the operation parameters and the noise characteristic data of the high-speed motor to obtain flux linkage real-time observation data; the self-adaptive analysis module is used for executing multi-mode working condition self-adaptive analysis according to the flux linkage real-time observation data and generating flux linkage control reference values of different rotating speed intervals; and the output module is used for inputting the flux linkage control reference values of different rotating speed intervals into a magnetic-machine coupling control algorithm, and calculating and outputting a motor low-noise driving control signal. According to the method, the problem of magnetic flux path distortion under high-speed operation is effectively solved, and downtime and efficiency loss caused by a traditional off-line calibration method can be avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of motor control, and particularly to a low-noise flux self-adaptive feedback control system for a high-speed motor. Background Art

[0002] When the motor speed reaches tens of thousands of revolutions per minute, traditional vector control and direct torque control methods face severe challenges. Traditional control methods usually adopt a single linear model, ignoring non-linear factors such as air-gap magnetic field distortion, eddy current effect, and cross saturation phenomenon under high-speed conditions, resulting in deviation in flux parameter identification, thus causing serious electromagnetic noise and mechanical vibration problems, affecting the equipment life and use comfort.

[0003] When a high-speed motor runs at high speed, the accuracy of its flux estimation will deteriorate severely with the changes in speed and temperature. Existing technologies mainly suppress noise by increasing the sampling frequency or the complexity of the filter, but these methods can only achieve limited effects under specific working conditions and cannot meet the control requirements of the full speed range. Especially in the case of sudden speed change, traditional flux control methods often lead to obvious whistling and vibration, affecting the system stability and control accuracy. In addition, existing control algorithms are difficult to simultaneously consider basic control performance and advanced optimization objectives, and cannot effectively suppress noise while ensuring the dynamic response of the system. Summary of the Invention

[0004] The main object of the present invention is to provide a low-noise flux self-adaptive feedback control system for a high-speed motor. The present invention effectively deals with the problem of flux path distortion under high-speed operation and can avoid the downtime and efficiency loss caused by traditional off-line calibration methods.

[0005] To achieve the above object, the present invention provides a low-noise flux self-adaptive feedback control system for a high-speed motor, including: An acquisition module, configured to acquire current, voltage, and vibration data of the high-speed motor at various operating speeds, and obtain high-speed motor operating parameters and noise characteristic data; A transformation processing module, configured to perform frequency-domain orthogonal transformation processing on the high-speed motor operating parameters and the noise characteristic data to obtain real-time flux observation data; An adaptive analysis module, configured to perform multi-modal working condition adaptive analysis according to the real-time flux observation data, and generate flux control reference values for different speed intervals; An output module, configured to input the flux control reference values for different speed intervals into a magnetic-mechanical coupling control algorithm, calculate and output a low-noise drive control signal for the motor.

[0006] Optionally, in the first implementation manner of the first aspect of the present invention, the acquisition module is specifically configured to: Obtain the rotational speed data of the high-speed motor, and divide the operating range of the high-speed motor into a low-speed area, a medium-speed area, and a high-speed area based on the rotational speed data; Inject frequency detection signals into the low-speed area, the medium-speed area, and the high-speed area respectively, and collect the three-phase voltage signals and three-phase current signals of the high-speed motor in the low-speed area, the medium-speed area, and the high-speed area; Perform Clarke transformation and Park transformation on the three-phase voltage signals to obtain d-q axis voltage data, and perform Clarke transformation and Park transformation on the three-phase current signals to obtain d-q axis current data; Collect the mechanical vibration signals and electromagnetic noise signals generated during the operation of the high-speed motor in the low-speed area, the medium-speed area, and the high-speed area, and perform fast Fourier transform on the mechanical vibration signals and the electromagnetic noise signals respectively to obtain noise characteristic data; Integrate the rotational speed data, the d-q axis voltage data, and the d-q axis current data to obtain the operating parameters of the high-speed motor.

[0007] Optionally, in the second implementation manner of the first aspect of the present invention, the transformation processing module further includes: A construction unit for constructing a frequency-domain orthogonal basis matrix including sine and cosine functions according to the rotational speed data in the operating parameters of the high-speed motor; An orthogonal transformation unit for performing orthogonal transformation on the d-q axis voltage data and the d-q axis current data by using the frequency-domain orthogonal basis matrix to obtain frequency-domain transformed electrical parameter data; A spectrum analysis unit for performing spectrum analysis on the noise characteristic data to obtain high-order harmonic components and slot harmonic components, and constructing a harmonic interference model according to the high-order harmonic components and the slot harmonic components; A filtering processing unit for performing harmonic noise identification and filtering processing on the frequency-domain transformed electrical parameter data based on the harmonic interference model to obtain filtered electrical parameters; A magnetic flux observation unit for calculating a magnetic flux coefficient matrix by using the filtered electrical parameters, and constructing an orthogonal magnetic flux observation model according to the magnetic flux coefficient matrix; A magnetic flux online calculation unit for obtaining the real-time voltage data and real-time current data of the high-speed motor, and inputting them into the orthogonal magnetic flux observation model for magnetic flux online calculation, and outputting magnetic flux real-time observation data.

[0008] Optionally, in the third implementation manner of the first aspect of the present invention, the magnetic flux observation unit is specifically used for: Convert the rotational speed data in the operating parameters of the high-speed motor into a first electrical angular frequency, and construct a frequency adaptive auxiliary variable matrix based on the first electrical angular frequency; Perform Fourier decomposition on the filtered electrical parameters to obtain a relationship curve between the flux linkage component and the frequency, and determine the harmonic order based on the relationship curve between the flux linkage component and the frequency; Construct an orthogonal basis function group according to the harmonic order and the frequency adaptive auxiliary variable matrix, and expand the flux linkage signal using the orthogonal basis function group to obtain a flux linkage orthogonal decomposition structure; Perform matrix multiplication on the filtered electrical parameters and the frequency adaptive auxiliary variable matrix to obtain an auxiliary variable covariance matrix; Add a regularization term to the auxiliary variable covariance matrix to obtain an enhanced covariance matrix, and solve the regularized least squares equation using the enhanced covariance matrix to obtain a flux linkage coefficient matrix; Construct a flux linkage observation equation according to the flux linkage coefficient matrix and the flux linkage orthogonal decomposition structure, and replace the time variable in the flux linkage observation equation with the phase angle calculated online to obtain an orthogonal flux linkage observation model.

[0009] Optionally, in the fourth implementation manner of the first aspect of the present invention, the flux linkage online calculation unit is specifically configured to: Collect the real-time voltage data and real-time current data of the high-speed motor, and perform phase compensation on the real-time voltage data and the real-time current data to obtain phase-compensated voltage and current data; Calculate a second electrical angular frequency according to the current rotational speed of the high-speed motor, and calculate the phase angle at the current moment based on the second electrical angular frequency to obtain the real-time phase information of the motor; Decompose the phase-compensated voltage and current data in the d-q coordinate system to obtain d-axis voltage and current components and q-axis voltage and current components; Update the flux linkage coefficient matrix according to the real-time phase information of the motor to obtain real-time auxiliary variable values; Substitute the real-time auxiliary variable values, the d-axis voltage and current components, and the q-axis voltage and current components into the orthogonal flux linkage observation model for flux linkage calculation to obtain real-time flux linkage observation data.

[0010] Optionally, in the fifth implementation manner of the first aspect of the present invention, the adaptive analysis module is specifically configured to: Classify and label the real-time flux linkage observation data to construct a low-speed area reference data set, a medium-speed area reference data set, and a high-speed area reference data set; Construct a multimodal reference model group based on the low-speed region reference data set, the medium-speed region reference data set, and the high-speed region reference data set. The multimodal reference model group includes a low-speed region flux linkage reference model, a medium-speed region flux linkage reference model, and a high-speed region flux linkage reference model; Perform a fast Fourier transform on the difference between the real-time flux linkage observation data and the output values of the multimodal reference model group to obtain a flux linkage observation error spectrum; Separate the flux linkage observation error spectrum into a low-frequency disturbance error and a parameter drift error, and perform integral compensation control and parameter adaptive law analysis on the low-frequency disturbance error and the parameter drift error to obtain a flux linkage error compensation amount; Calculate the weight coefficients of the low-speed region flux linkage reference model, the medium-speed region flux linkage reference model, and the high-speed region flux linkage reference model at the current speed and temperature, and perform weighted fusion based on the weight coefficients to obtain initial flux linkage reference values for different speed intervals; Superimpose the initial flux linkage reference values for different speed intervals with the flux linkage error compensation amount respectively to generate flux linkage control reference values for different speed intervals.

[0011] Optionally, in the sixth implementation manner of the first aspect of the present invention, the output module is specifically configured to: Decompose the flux linkage control reference values for different speed intervals to obtain a linear inductance matrix part and a non-linear flux linkage compensation part; Execute current control based on the linear inductance matrix part to obtain a basic control quantity; Input the current current vector and speed data of the high-speed motor into a magnetic-mechanical coupling control algorithm for calculation to obtain a non-linear flux linkage compensation control quantity; Execute control signal synthesis based on the basic control quantity and the non-linear flux linkage compensation control quantity to obtain a synthesized control signal; Perform harmonic interference identification and compensation on the synthesized control signal to obtain a compensated control signal, and perform pulse width modulation transformation on the compensated control signal to output a low-noise drive control signal for the motor.

[0012] Optionally, in the seventh implementation manner of the first aspect of the present invention, the low-noise flux linkage adaptive feedback control system of the high-speed motor further includes: The hierarchical control module is used to distribute the low-noise drive control signal of the motor to the basic control layer, the adaptive estimation layer and the intelligent optimization layer to obtain hierarchical control instructions; based on the noise characteristic data of the high-speed motor, different execution priorities and control bandwidths are set for the hierarchical control instructions to obtain a priority control strategy; the basic adjustment of the motor current is performed according to the control instructions of the basic control layer in the hierarchical control instructions to obtain a basic current adjustment amount; the dynamic compensation is performed on the basic current adjustment amount based on the control instructions of the adaptive estimation layer in the hierarchical control instructions to obtain a compensated current control amount; the sudden change working conditions of the high-speed motor are smoothed by using the control instructions of the intelligent optimization layer in the hierarchical control instructions to generate a smooth transition control signal; the compensated current control amount and the smooth transition control signal are coordinately synthesized and processed with a dual sampling rate to output a full-speed range low-noise current control instruction.

[0013] In summary, the technical solution provided by the present invention effectively filters high-frequency interference and noise through the frequency adaptive auxiliary variable and the orthogonal flux observer, suppresses the propagation of electromagnetic noise from the source, and greatly reduces the electromagnetic noise and mechanical vibration during the operation of the high-speed motor, improving the operation comfort and stability of the equipment. The multi-modal model reference adaptive system is adopted to divide the motor working range into multiple modes and achieve smooth transition. Combining the spectrum analysis mechanism of the flux observer error, the errors caused by load disturbance and parameter drift are distinguished, and high-precision flux control performance is maintained within the full-speed range. The non-linear flux model is decomposed through special coordinate transformation, and combined with the adaptive high-order neural network structure, the complex non-linear relationship of the flux in the weak magnetic region of the high-speed motor is accurately modeled, effectively dealing with the problem of flux path distortion under high-speed operation. The smooth transition mechanism is designed to ensure smooth transition of the control output during large-scale speed changes, avoid control oscillation, significantly improve the stability and comfort of the system during rapid speed changes, and reduce speed fluctuations and abnormal noises. The real-time self-calibration mechanism combined with the working condition identifier can maintain the best performance during long-term operation and various working conditions, automatically select the optimization strategy suitable for the current working condition, and maintain consistent control performance under extreme temperature and complex load conditions. The dual sampling rate control strategy and the hierarchical control architecture are adopted. The low sampling rate subsystem is used for the basic control layer and the adaptive estimation layer, and the high sampling rate subsystem is used for the intelligent optimization layer and spectrum analysis, optimizing the system resource allocation, reducing the control delay, and improving the overall control efficiency. The real-time self-calibration mechanism is designed to complete parameter update during normal operation by injecting detection signals with specific frequencies and detecting response characteristics, improving the continuous availability of the system and avoiding the downtime and efficiency loss caused by traditional offline calibration methods. Description of the Drawings

[0014] Figure 1 It is a structural block diagram of a low-noise flux adaptive feedback control system for a high-speed motor in an embodiment of the present invention.

[0015] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] Reference Figure 1 , this embodiment provides a low-noise flux adaptive feedback control system for a high-speed motor, comprising: The acquisition module 11 is used to collect the current, voltage and vibration data of the high-speed motor at various operating speeds to obtain the operating parameters and noise characteristic data of the high-speed motor; Among them, by obtaining the speed data of the high-speed motor, the motor operating state is divided according to the speed range. The range division adopts the method of low-speed zone, medium-speed zone and high-speed zone to ensure accurate adaptation of the dynamic characteristics of the motor in the whole life cycle and all working conditions. According to the real-time detected speed information, the current speed range is dynamically determined, and the parameters and strategies for different ranges are set accordingly to deal with the magnetic flux characteristics, loss changes and noise-dominant mechanisms of the high-speed motor in different speed ranges in a targeted manner. In order to improve the comprehensive perception of the motor operating state, the designed frequency detection signals are injected into the system in the low-speed zone, medium-speed zone and high-speed zone respectively. These signals can induce the motor to respond to the dynamic and spectral characteristics unique to each range. While injecting the frequency detection signal, the system synchronously collects the three-phase voltage signal and the three-phase current signal of the motor with high precision. The three-phase voltage and current signals are subjected to Clarke transformation and Park transformation. The Clarke transformation maps the three-phase AC signal to the α-β component in the two-phase stationary coordinate system, simplifies the analysis complexity of the three-phase signal, and then the Park transformation rotates the α-β component to the dq synchronous rotating coordinate system. Through these two steps of coordinate transformation, the dq axis voltage data and dq axis current data are obtained. At the same time, through the collaborative work of multi-channel sensors and high-frequency data acquisition cards, the mechanical vibration signals and electromagnetic noise signals of the motor in each speed range are synchronously collected. In order to reveal the spectral structure and time domain evolution characteristics of the noise signal, fast Fourier transform is performed on all collected vibration signals and electromagnetic noise signals, and the time domain signal is decomposed into the amplitude distribution of each frequency component to obtain full spectrum characteristic data reflecting the dual mechanical and electromagnetic noise mechanisms. The speed data, voltage and current data after dq axis transformation, and characteristic data of vibration noise are integrated to form a set of high-speed motor operation parameters containing multi-modal and multi-dimensional parameters.

[0018] A transformation processing module 12 is configured to perform frequency-domain orthogonal transformation processing on the operating parameters and noise characteristic data of the high-speed motor to obtain real-time magnetic flux observation data; Specifically, with the rotational speed data as the core variable, a frequency-domain orthogonal basis matrix adapted to the current operating condition of the motor is dynamically constructed. According to the real-time collected rotational speed information, the current electrical angular frequency of the motor is calculated, and based on this, an orthogonal basis group with sine and cosine functions as the core elements is generated. This basis group includes the basic sin(ωt) and cos(ωt), and is extended to higher harmonic forms such as sin(nωt) and cos(nωt), so as to completely cover the fundamental frequency, harmonics and their related spectral characteristics of the motor in the full speed range in the frequency domain. The orthogonal transformation unit uses the above-mentioned frequency-domain orthogonal basis matrix to perform orthogonal transformation processing on the d-q axis voltage data and the d-q axis current data. That is, on an adaptive frequency-domain basis, the instantaneous electrical parameters of the motor are mapped to each orthogonal component in the frequency space, thereby separating the contributions of different frequency components to the overall magnetic flux of the motor, and obtaining the electrical parameter data after frequency-domain transformation. The spectrum analysis unit performs spectrum analysis on the noise characteristic data. By spectrum analysis methods such as fast Fourier transform, the high-order harmonic components and slot harmonic components contained in the electrical noise are identified, and a harmonic interference model is established accordingly. This model records the amplitude, phase and frequency distribution of each type of main harmonic, and reflects the harmonic dominant characteristics of the motor in different speed ranges in real time. The filtering processing unit then performs harmonic noise identification and filtering on the above-mentioned electrical parameter data after frequency-domain transformation according to the established harmonic interference model. This unit adopts a variety of filtering strategies such as adaptive high-pass, notch, and band-stop, and dynamically adjusts the filtering parameters according to the actual operating state of the motor, effectively minimizing the negative impact of high-order harmonics and slot harmonics on the magnetic flux observation accuracy, and obtaining the filtered electrical parameters. Based on the filtered electrical parameters, the magnetic flux observation unit calculates a magnetic flux coefficient matrix according to the magnetic flux analysis theory of the high-speed motor, in combination with the electrical parameters and the motor structure parameters. This coefficient matrix, as the core parameter of the orthogonal magnetic flux observation model, can dynamically construct a high-precision magnetic flux estimation model in real time according to the dynamic changes of the electrical parameters under different operating conditions. During the operation of the system, the magnetic flux online calculation unit continuously obtains the real-time voltage and current data of the motor, and uses this as the input to drive the orthogonal magnetic flux observation model to complete the online estimation of the motor magnetic flux, and obtain the real-time magnetic flux observation data.

[0019] In this embodiment, the rotational speed data in the operating parameters of the high-speed motor is converted into the first electrical angular frequency, which reflects the rotational speed of the current magnetic field of the motor. Based on this electrical angular frequency, a frequency adaptive auxiliary variable matrix is constructed. This auxiliary variable matrix consists of several sine and cosine functions that vary dynamically with frequency, such as [sin(ωt), cos(ωt), sin(3ωt), cos(3ωt),...], ensuring that the auxiliary variables are always highly matched with the actual operating state of the motor throughout the rotational speed range and providing an orthogonal mathematical basis for subsequent signal decomposition and feature extraction. After the construction of the auxiliary variable matrix, Fourier decomposition is performed on the electrical parameters (such as d-q axis current and voltage) that have been filtered. Fourier decomposition can decompose the electrical signals in the time domain into the superposition of several different frequency components, obtaining the relationship curve between the flux linkage components and the frequency. This relationship curve reveals the variation law of the main flux linkage with different frequency components, and reflects the existence and intensity of the harmonic components, facilitating the system to accurately determine the harmonic order participating in the flux linkage decomposition according to the curve characteristics. According to the harmonic order and the frequency adaptive auxiliary variable matrix, a set of orthogonal basis function groups are constructed. These basis functions cover all significant harmonic components and have the adaptive ability of frequency tracking. Using the orthogonal basis function groups, the flux linkage signal is expanded to obtain the orthogonal decomposition structure of the flux linkage, that is, the original complex flux linkage time domain signal is accurately expressed by a set of orthogonal components with variable amplitudes and phases, thereby improving the resolution and robustness of the flux linkage observation. The filtered electrical parameters are multiplied by the frequency adaptive auxiliary variable matrix through matrix multiplication to obtain the auxiliary variable covariance matrix that reflects the contribution of different auxiliary variable components. The covariance matrix reveals the correlation and linear combination relationship between the auxiliary variables. To prevent ill-conditioned problems caused by signal similarity or noise interference, a regularization term is introduced based on the covariance matrix, and the main diagonal elements are enhanced through the weighted penalty parameter to form an enhanced covariance matrix. Based on the enhanced covariance matrix, the regularized least squares equation is used for solution to obtain the coefficient matrix of the flux linkage, which reflects the weight and physical meaning of each orthogonal basis function component in the overall flux linkage signal. Combining the flux linkage coefficient matrix with the orthogonal decomposition structure of the flux linkage, a flux linkage observation equation is constructed to achieve accurate modeling of the complex flux linkage dynamics. To highly integrate the theoretical model with the actual operating conditions, the time variable in the flux linkage observation equation is replaced with the phase angle information calculated in real time to form an orthogonal flux linkage observation model based on phase drive.

[0020] In this embodiment, the real-time voltage data and real-time current data of the high-speed motor are collected, and phase compensation is performed on the real-time voltage data and real-time current data to eliminate the phase shift and error introduced in the signal acquisition process due to links such as sensor response, data transmission delay, and system filtering, so as to ensure the physical alignment and time synchronization of voltage and current in the motor rotating coordinate system in subsequent calculations, and obtain the voltage and current data after phase compensation. Using the currently acquired high-speed motor speed information, the second electrical angular frequency is dynamically calculated. The second electrical angular frequency is the product of the real-time motor speed and the number of pole pairs, and determines the time-varying characteristics of the motor electromagnetic state. On this basis, integration is performed on the current moment based on this frequency to obtain the current phase angle, and the phase information reflecting the real-time electrical angle of the motor is obtained. The voltage and current data after phase compensation processing are input into the coordinate transformation module, and the system expands according to the d-q coordinate system, and transforms the signals in the three-phase stationary coordinates into the d-axis voltage and current components and q-axis voltage and current components that rotate synchronously with the rotor magnetic field. This step can effectively separate the flux linkage control and energy flow, so that the d-axis component mainly represents the excitation and main flux linkage, while the q-axis component reflects the output torque and power. Based on the real-time phase information of the motor, the flux linkage coefficient matrix is dynamically updated. The flux linkage coefficient matrix reflects the weight relationship of voltage and current signals on different orthogonal auxiliary variable components, and the activation and weight distribution of these components need to be adjusted in a timely manner according to the continuous change of the motor phase. By combining the current phase information with a preset or adaptive algorithm, the system calculates the real-time auxiliary variable value, so that each orthogonal component can highly match the time-varying characteristics of the motor electromagnetic state, and enhances the sensitivity and response ability of the observer to the change of working conditions. The above real-time auxiliary variable value and the decomposed d-axis and q-axis voltage and current components are jointly input into the orthogonal flux linkage observation model, and through the dynamic solution of the model, the real-time observation data of the flux linkage are obtained.

[0021] The adaptive analysis module 13 is used to perform multi-modal working condition adaptive analysis according to the real-time flux linkage observation data and generate the flux linkage control reference values for different speed intervals; It should be noted that, with the real-time observed flux linkage data as the core, by continuously collecting and automatically classifying and labeling all the flux linkage observation values during the operation of the motor, the data is divided into a low-speed area, a medium-speed area, and a high-speed area according to the preset speed range, and a reference data set for the low-speed area, a reference data set for the medium-speed area, and a reference data set for the high-speed area covering the entire life cycle are respectively constructed. Based on these three partition reference data sets, methods such as machine learning modeling, statistical regression, and system identification are used to train a low-speed area flux linkage reference model, a medium-speed area flux linkage reference model, and a high-speed area flux linkage reference model respectively. These models form a multi-modal reference model group for different operating conditions, which specifically describes the ideal evolution trajectory, energy distribution, and nonlinear change characteristics of the flux linkage of the motor in each speed range. During the actual operation process, the system continuously compares and analyzes the real-time observed flux linkage data with the output of the multi-modal reference model group under the corresponding operating conditions, and performs spectral decomposition on the observation error between the two through fast Fourier transform to obtain the amplitude and characteristic distribution of the observation error in different frequency ranges. This error spectrum contains both low-frequency disturbance components caused by load disturbances, external shocks, etc., and high-frequency or slow-varying parameter drift components caused by motor parameter drift, environmental changes, etc. The error spectrum is separated and extracted to distinguish the low-frequency disturbance error and the parameter drift error, and differential compensation and adaptive analysis methods are adopted. For the low-frequency disturbance error, an integral compensation control mechanism is introduced. By accumulating the error and implementing feedback adjustment, the steady-state offset caused by low-frequency disturbances such as load fluctuations and resonance disturbances is effectively suppressed; for the parameter drift error, based on the principle of adaptive law, the core parameters of the multi-modal reference model are dynamically corrected to achieve adaptive tracking and automatic compensation for long-term slow-changing factors such as temperature changes and material aging. On the basis of error compensation, considering the actual speed and environmental temperature of the current motor, the weight coefficients of the low-speed area, medium-speed area, and high-speed area flux linkage reference models are dynamically calculated. The weight coefficients adopt Gaussian or exponential weight functions to ensure smooth transition during operating condition switching and critical intervals, and avoid control jitter and discontinuity phenomena in the traditional hard switching method. By weighted fusion of the output results of the multi-modal flux linkage reference model under different weights, an initial flux linkage reference value reflecting the current operating condition characteristics is generated in real time. The initial flux linkage reference values in the above-mentioned respective speed ranges are superimposed one by one with the corresponding flux linkage error compensation amounts to generate flux linkage control reference values with adaptive ability, higher accuracy, and faster dynamic response in different speed ranges.

[0022] An output module 14 is used to input the flux linkage control reference values in different speed ranges into the magnetic-mechanical coupling control algorithm, calculate and output a low-noise drive control signal for the motor.

[0023] Specifically, the flux control reference values in different rotational speed ranges generated by the front-end adaptive analysis module are structurally decomposed. The electromagnetic dynamics information implicit in the reference values is divided into a linear inductance matrix part and a non-linear flux compensation part. The linear inductance matrix part mainly reflects the direct and proportional relationship between current and flux under the assumption of an ideal linear magnetic field in high-speed motors; while the non-linear flux compensation part is a dynamic deviation term actively extracted and compensated by the system when the flux is disturbed by non-ideal factors such as material saturation, hysteresis, eddy current effects, and higher harmonics under high-speed operating conditions. Based on this distinction, the traditional current loop control is executed using the linear inductance matrix part. Through PI or other linear adjustment algorithms, the required basic current control quantity is calculated in real time to ensure that the motor obtains the expected main flux amplitude and steady-state output under normal excitation conditions. At the same time, the current vector and real-time rotational speed data of the current high-speed motor are input into the magnetic-mechanical coupling control algorithm module. This module relies on neural networks or adaptive modeling techniques to dynamically calculate the multi-dimensional compensation quantity for the non-linear flux relationship. This process can actively identify the weak magnetic region, strong magnetic region, and the non-linear change law of the flux under different operating states, effectively making up for the adjustment lag and flux fluctuation that occur when only relying on linear model control, and improving the control accuracy and dynamic performance of the entire system under extreme operating conditions. Based on the basic control quantity and the non-linear flux compensation control quantity, the control signal synthesis is executed to obtain the synthesized control signal. The harmonic interference in the synthesized control signal is identified and compensated. Based on the aforementioned frequency domain observation and harmonic model, the high-order harmonics, slot harmonics, and specific noise frequency components of the synthesized signal are detected and separated in real time. For the detected significant harmonic components, digital processing algorithms such as notch, band-stop, and frequency adaptive filtering are used to implement active compensation, thereby suppressing the high-frequency noise and vibration jointly caused by the control loop, the motor body, and external disturbances to the greatest extent. The control signal after harmonic compensation is input into the pulse width modulation transformation module. By dynamically and adaptively adjusting parameters such as the carrier frequency and modulation ratio, the precise generation of the inverter gate drive signal is achieved, and the low-noise drive control signal for the motor is output.

[0024] Structurally allocate the generated low-noise drive control signal for the motor. The allocation process is based on the hierarchical idea of control theory, dividing the entire control task into three major functional modules: the basic control layer, the adaptive estimation layer, and the intelligent optimization layer. Each layer undertakes control responsibilities with different granularities and focuses. The overall control signal is logically parsed and distributed according to strategies, and then transmitted to these three control layers respectively to generate hierarchical control instructions that complement each other and are closely linked. The basic control layer focuses on achieving fast and stable regulation of basic physical quantities such as motor current, ensuring the real-time performance and stability of the system in terms of main frequency and steady-state indicators. The adaptive estimation layer is mainly responsible for modeling and adaptive compensation of uncertainties caused by magnetic flux, parameter perturbations, and operating condition changes, thereby enhancing the system's adaptability to parameter fluctuations and external disturbances. The intelligent optimization layer stands at a higher strategic level. Based on real-time operating condition monitoring and historical operation big data, it dynamically adjusts the weights and priorities of control algorithms, and performs smooth transition optimization for state switches of high-speed motors under sudden operating conditions such as rapid acceleration and rapid deceleration, minimizing system oscillations and the excitation of high-frequency noise. Based on the noise characteristic data collected from the high-speed motor, the hierarchical control module not only assigns different execution priorities to the hierarchical control instructions, but also allocates different bandwidth resources to each layer of control in combination with the actual noise spectrum characteristics. For example, the basic control layer gives priority to ensuring large bandwidth and low latency, focusing on the high-speed response of the current loop, while the adaptive estimation layer sacrifices the response speed appropriately and focuses on improving the robustness of disturbance identification and parameter update. The intelligent optimization layer aims at strategy smoothing and global performance optimization. By dynamically switching bandwidth and priority, it achieves multi-objective trade-off and optimal allocation, constructing a dynamic and flexible priority control strategy system. In the specific control process, the hierarchical control module completes the basic regulation of the motor current based on the instructions issued by the basic control layer, obtaining a high-bandwidth basic current regulation amount to ensure the active response of the drive. The adaptive estimation layer dynamically compensates the above basic current regulation amount according to the hierarchical control instructions it receives, identifies and corrects the deviations caused by model uncertainties, external load changes, or operating environment impacts, and obtains a more accurate and robust current control amount after compensation. When sudden states occur in the system, such as large changes in speed or sudden load changes, the intelligent optimization layer smooths the transition process of the entire system according to its strategy instructions, generating a smooth transition control signal to suppress system oscillations and high-frequency noise caused by drastic operating condition switches, so that the control output can still maintain a highly smooth and orderly dynamic response when dealing with extreme operating conditions. To ensure the efficient cooperation and resource utilization among hierarchical control signals, the hierarchical control module designs a coordination synthesis mechanism and a dual-sampling rate processing strategy. The current control amount after compensation and the smooth transition control signal will undergo weight fusion and logical coordination at this stage.The system robustness and computational efficiency are achieved by the basic control layer and the adaptive estimation layer with a low sampling rate. The intelligent optimization layer and the transition control track the sudden change of working conditions at a high sampling rate, and the three cooperate to output the final full-speed low-noise current control command.

[0025] In one embodiment, the acquisition module 11 is specifically configured to: Obtain the speed data of the high-speed motor, and divide the working range of the high-speed motor into a low-speed area, a medium-speed area, and a high-speed area based on the speed data; Inject frequency detection signals in the low-speed area, the medium-speed area, and the high-speed area respectively, and collect the three-phase voltage signals and three-phase current signals of the high-speed motor in the low-speed area, the medium-speed area, and the high-speed area; Perform Clarke transformation and Park transformation on the three-phase voltage signals to obtain d-q axis voltage data, and perform Clarke transformation and Park transformation on the three-phase current signals to obtain d-q axis current data; Collect the mechanical vibration signals and electromagnetic noise signals generated during the operation of the high-speed motor in the low-speed area, the medium-speed area, and the high-speed area, and perform fast Fourier transform on the mechanical vibration signals and electromagnetic noise signals respectively to obtain noise characteristic data; Integrate the speed data, d-q axis voltage data, and d-q axis current data to obtain the operating parameters of the high-speed motor.

[0026] Among them, the rotational speed data of the high-speed motor is continuously and real-time collected through a rotational speed sensor, and the accuracy and stability of the rotational speed signal are improved through digital filtering and denoising algorithms. With the accumulation of the data stream, the system sets interval thresholds based on the rotational speed data as the core basis, combined with the motor design specifications and the actual application requirements, and divides the full-speed operation range of the motor into a low-speed area, a medium-speed area, and a high-speed area. The low-speed area is used for delicate control scenarios such as starting and low-load, the medium-speed area covers the daily main working conditions, and the high-speed area covers complex working conditions such as high-speed operation and field-weakening speed increase. In each rotational speed interval, the active signal injection method is adopted, and frequency detection signals designed specifically are injected into the low-speed area, the medium-speed area, and the high-speed area respectively through a driver or a dedicated signal source. These detection signals are sinusoidal waves, square waves, or pulse sequences with controlled amplitudes, and their frequencies, amplitudes, and injection methods are set according to the working condition characteristics of their respective intervals. By injecting the frequency detection signals, the intrinsic dynamic responses of the motor in different speed intervals are excited, and the identification ability of harmonics, losses, and field-weakening phenomena in the motor system is enhanced. At the same time, the acquisition system synchronously acquires the three-phase voltage signals and three-phase current signals of the motor in the three major speed intervals at a high sampling rate, and ensures the integrity and high fidelity of signal acquisition through a low-latency analog front end and a high-resolution ADC. For the collected three-phase signals, in order to extract the intrinsic characteristics of the motor operation mechanism, the Clarke transformation and the Park transformation are sequentially performed on the three-phase voltage and three-phase current signals. The Clarke transformation maps the AC signals in the three-phase stationary coordinate system into the orthogonal components of the α-β bivariate, reducing the redundancy and complexity of the signals, while the Park transformation rotates and transforms the α-β components into the d-q coordinate components that rotate synchronously with the magnetic field according to the real-time rotational speed and electrical angular frequency of the motor. After processing, the originally complex three-phase signals are converted into the d-axis and q-axis voltage and current data that intuitively reflect the essence of flux linkage and torque control, where the d-axis component directly describes the excitation component and the change of the main flux linkage, and the q-axis component corresponds to the output torque and the dynamic energy conversion. At the same time, the mechanical vibration signals and electromagnetic noise signals during the operation of the high-speed motor are collected in parallel in each rotational speed interval. The mechanical vibration signals are collected in real time through a high-sensitivity acceleration sensor or a laser vibrometer, while the electromagnetic noise signals are obtained through a distributed microphone array or a structure-integrated electro-acoustic sensor. All the collected mechanical and electromagnetic signals are quickly cached in the time domain, and then uniformly input into the signal processing module to perform spectrum analysis algorithms such as the fast Fourier transform. The fast Fourier transform can decompose the time-domain vibration and noise signals into various order frequency components and amplitude spectral lines, revealing complex noise mechanisms such as mechanical resonance, bearing faults, slot harmonics, and electromagnetic excitation in different speed intervals. By analyzing the noise characteristic data, the main noise sources are traced and located, and it can provide a basis for subsequent harmonic suppression, adaptive filtering, and structure optimization strategies, realizing the full-link closed-loop management of the motor system noise from the source to the structure.Integrate the rotational speed data, d-q axis voltage data, and d-q axis current data obtained from the above acquisition and processing, as well as the noise characteristic data obtained from the spectrum analysis of vibration and noise signals, to construct a multi-modal and full-dimensional set of high-speed motor operating parameters.

[0027] In one embodiment, the transformation processing module 12 further includes: A construction unit for constructing a frequency-domain orthogonal basis matrix including sine and cosine functions according to the rotational speed data in the high-speed motor operating parameters; An orthogonal transformation unit for performing orthogonal transformation on the d-q axis voltage data and d-q axis current data using the frequency-domain orthogonal basis matrix to obtain the electrically parameter data after frequency-domain transformation; A spectrum analysis unit for performing spectrum analysis on the noise characteristic data to obtain high-order harmonic components and slot harmonic components, and constructing a harmonic interference model based on the high-order harmonic components and slot harmonic components; A filtering processing unit for performing harmonic noise identification and filtering processing on the electrically parameter data after frequency-domain transformation based on the harmonic interference model to obtain the filtered electrically parameters; A flux linkage observation unit for calculating a flux linkage coefficient matrix using the filtered electrically parameters and constructing an orthogonal flux linkage observation model based on the flux linkage coefficient matrix; A flux linkage online calculation unit for acquiring the real-time voltage data and real-time current data of the high-speed motor, inputting them into the orthogonal flux linkage observation model for online flux linkage calculation, and outputting the real-time flux linkage observation data.

[0028] Specifically, based on the rotational speed data, the current electrical angular frequency is dynamically calculated and used as the main reference for frequency-domain analysis. On this basis, the construction unit generates a series of orthogonal basis functions according to the electrical angular frequency. These functions include sin(ωt), cos(ωt), and higher-order sin(nωt), cos(nωt) (where n is the harmonic order). They are arranged and combined to form an adaptive frequency-domain orthogonal basis matrix. The construction method of this basis matrix ensures that regardless of the current rotational speed range of the motor, the mathematical basis used can accurately match the operating state of the motor itself, realizing signal projection and decomposition in the full speed range and multiple frequency bands. The orthogonal transformation unit performs signal mapping on the collected d-q axis voltage data and d-q axis current data. Using the frequency-domain orthogonal basis matrix constructed in the previous step, this unit decomposes the time-domain voltage and current signals onto these orthogonal bases by means of mathematical methods such as projection, inner product, or least squares fitting, realizing the conversion from the time domain to the frequency domain. Through this step, the original complex signal containing various periodic components and instantaneous changes is transformed into frequency-domain parameter data characterized by the main frequency and each order of harmonics. In the frequency domain, each component represents the specific contribution of the corresponding fundamental frequency or harmonic component to the overall signal. The spectrum analysis unit deeply deconstructs the noise characteristic data of the entire system. Through spectrum analysis algorithms such as the fast Fourier transform, this unit extracts the high-order harmonic components and slot harmonic components in the mechanical vibration and electromagnetic noise signals. These components are the main components of the motor system noise, originating from the resonance of the structural natural frequency and excitation, or being excited by factors such as uneven stator winding layout, magnetic field distortion, inverter switching, and load disturbance. The system not only identifies these harmonic components, further analyzes their amplitudes, phases, and energy distributions, but also comprehensively constructs a harmonic interference model that fits the actual operating environment. This model records the frequencies, energies of the dominant noise components and their mapping relationships with the motor operating state. Based on the harmonic interference model, the filtering processing unit initiates targeted harmonic noise identification and filtering strategies. The system dynamically determines which frequency bands in the current signal belong to the main harmonic noise area, and accordingly configures algorithm modules such as notch, band-stop, high-pass, or adaptive digital filters to specifically suppress high-order harmonics, slot harmonics, and other identifiable noise signals. During the filtering process, it is ensured that effective information (such as the main magnetic flux linkage, fundamental wave components, etc.) can pass through the control loop with high fidelity, while maximizing the real-time attenuation of disturbance and useless noise components, and obtaining the filtered electrical parameters. The magnetic flux linkage observation unit takes the filtered electrical parameters as input and calculates the magnetic flux linkage coefficient matrix online according to the equivalent electromagnetic model and real-time operating parameters of the high-speed motor. This coefficient matrix is the result of comprehensively encoding multiple factors such as the dynamic relationship between voltage and current in the d-q coordinate system, the influence of rotational speed, motor structure parameters, frequency response, and harmonic interference from both physical and engineering perspectives. With the help of this coefficient matrix, a set of orthogonal magnetic flux linkage observation models are constructed to map all basic variables and parameters into real-time estimated values of the main magnetic flux linkage and each order of harmonic components.The core advantage of the orthogonal flux linkage observation model lies in its use of the frequency-domain orthogonal characteristics and adaptive coefficients to achieve robust tracking of complex operating conditions of motors, such as nonlinearity, wide frequency band, high-speed weak magnetic field, etc. For both fundamental and harmonic components, stable and high-resolution real-time observation results can be obtained. The flux linkage online calculation unit performs a full-process closed-loop real-time calculation. This unit continuously obtains the real-time voltage data and real-time current data of the motor, and at the same time uses the aforementioned orthogonal flux linkage observation model to project and decompose these signals online to accurately reconstruct the instantaneous main flux linkage and harmonic flux linkage. The observation results are fed back to the control system as the input basis for subsequent adaptive adjustment, and are output to peripheral systems such as state monitoring, health diagnosis, and performance analysis to achieve self-sensing, self-adjustment, and self-optimization of the motor in multiple dimensions and under all operating conditions.

[0029] In one embodiment, the flux linkage observation unit is specifically used for: Convert the rotational speed data in the operating parameters of the high-speed motor into the first electrical angular frequency, and construct a frequency-adaptive auxiliary variable matrix based on the first electrical angular frequency; Perform Fourier decomposition on the filtered electrical parameters to obtain the relationship curve between the flux linkage component and the frequency, and determine the harmonic order based on the relationship curve between the flux linkage component and the frequency; Construct an orthogonal basis function group according to the harmonic order and the frequency-adaptive auxiliary variable matrix, and expand the flux linkage signal using the orthogonal basis function group to obtain the orthogonal decomposition structure of the flux linkage; Perform matrix multiplication on the filtered electrical parameters and the frequency-adaptive auxiliary variable matrix to obtain the auxiliary variable covariance matrix; Add a regularization term to the auxiliary variable covariance matrix to obtain an enhanced covariance matrix, and use the enhanced covariance matrix to solve the regularized least squares equation to obtain the flux linkage coefficient matrix; Construct a flux linkage observation equation according to the flux linkage coefficient matrix and the orthogonal decomposition structure of the flux linkage, and replace the time variable in the flux linkage observation equation with the phase angle calculated online to obtain the orthogonal flux linkage observation model.

[0030] In this embodiment, the rotational speed data in the operating parameters of the high-speed motor is converted into the first electrical angular frequency. The calculation of the electrical angular frequency involves the product of the current rotational speed of the motor and the number of pole pairs. The electrical angular frequency determines the rotational speed of the motor electromagnetic field and directly affects the main frequencies of periodic components such as sine and cosine in the system. It is the reference variable for frequency-domain modeling and signal decomposition. With this electrical angular frequency as the core, a set of frequency adaptive auxiliary variable matrices are constructed based on the principle of mathematical modeling. This matrix uses sine and cosine functions as basic components and is extended to high-order harmonic forms according to actual needs, such as sin(ωt), cos(ωt), sin(3ωt), cos(3ωt), etc. Through the design of adaptive auxiliary variables, the matrix can adjust its own structure in real time with the changes of rotational speed and frequency, so that it always highly matches the current operating state of the motor. The filtered electrical parameters (such as d-q axis current and voltage signals) are input into the Fourier decomposition unit. Through Fourier decomposition, the electrical signals in the time domain are restored to the superposition of multiple different frequency components, and each component corresponds to different flux linkage dynamic characteristics. The system thus obtains the relationship curve between the flux linkage components and the frequency, reflecting the energy distribution of different frequency components (such as fundamental wave, each order harmonic) in the total flux linkage signal. By analyzing the relationship curve, identify which frequency components dominate under the current operating conditions, and quantitatively judge the energy ratio between the high-order harmonics and the fundamental wave. By analyzing the energy distribution and attenuation trend of each frequency component, adaptively determine the effective harmonic order, that is, the number of high-order harmonic components to be considered in the actual orthogonal decomposition. Based on the adaptively obtained harmonic order, the system combines the constructed frequency adaptive auxiliary variable matrix to generate an orthogonal basis function group. When the orthogonal basis function is applied to the expansion of the flux linkage signal, the orthogonal decomposition of the original flux linkage signal is realized, and the complex, non-stationary, and non-linear superimposed time-domain signal is restored to the linear combination of several orthogonal components, obtaining the flux linkage orthogonal decomposition structure. Perform matrix multiplication on the filtered electrical parameters and the frequency adaptive auxiliary variable matrix to calculate the correlation and collaborative contribution of different auxiliary variable components in the overall signal, and generate an auxiliary variable covariance matrix. The covariance matrix records the statistical correlation between each pair of auxiliary variables. Introduce a regularization term on the covariance matrix, which is manifested as adding a small positive number on the main diagonal to obtain an enhanced covariance matrix. Regularization effectively suppresses the ill-posedness of the solution space caused by signal noise or insufficient data, and improves the numerical stability and robustness of parameter identification. Using the enhanced covariance matrix, perform parameter calculation through the regularized least squares equation to obtain the flux linkage coefficient matrix. The goal of the least squares method is to find a set of optimal weights to minimize the sum of the squares of the errors between the flux linkage signal reconstructed by the orthogonal decomposition structure and the actual observed signal, and the regularization term further restricts the complexity of the solution and the abnormal fluctuations of the parameters. Each item in the flux linkage coefficient matrix represents the actual contribution of a certain orthogonal basis component in the flux linkage signal. Combining the coefficients of all components, the original flux linkage time-domain signal is restored.Construct a flux linkage observation equation based on the flux linkage coefficient matrix and the orthogonal decomposition structure of the flux linkage. This equation describes the dynamic evolution relationship of the flux linkage under the current working conditions and signal structure in the form of weighted summation of orthogonal basis components and coefficients. To improve the real-time performance and physical consistency of the model, the time variable in the flux linkage observation equation is replaced with the online phase angle obtained by integrating the high-speed motor speed and the number of pole pairs, so that the time-varying characteristics of the orthogonal basis are highly synchronized with the actual motor electromagnetic state and dynamically adapt to the drastic changes in the motor speed, resulting in an orthogonal flux linkage observation model.

[0031] In one embodiment, the flux linkage online calculation unit is specifically configured to: Collect the real-time voltage data and real-time current data of the high-speed motor, and perform phase compensation on the real-time voltage data and real-time current data to obtain the voltage and current data after phase compensation; Calculate the second electrical angular frequency according to the current speed of the high-speed motor, and calculate the phase angle at the current moment based on the second electrical angular frequency to obtain the real-time phase information of the motor; Decompose the voltage and current data after phase compensation according to the d-q coordinate system to obtain the d-axis voltage and current components and the q-axis voltage and current components; Update the flux linkage coefficient matrix according to the real-time phase information of the motor to obtain the real-time auxiliary variable value; Substitute the real-time auxiliary variable value, the d-axis voltage and current components, and the q-axis voltage and current components into the orthogonal flux linkage observation model for flux linkage calculation to obtain the real-time observed data of the flux linkage.

[0032] In this embodiment, the real-time voltage data and real-time current data of the high-speed motor are collected, and multi-channel synchronous sampling is used to ensure the time alignment of the data in each channel at the moment of collection, so as to avoid the negative impact of physical timing errors on subsequent modeling and control. Phase compensation is performed on all the collected voltage and current signals, including various means such as frequency-domain phase correction, time-axis alignment, interpolation compensation, or system modeling back-calculation, to restore the original signal to the physical reference time point and obtain the voltage and current data after phase compensation. On the basis data set with consistent phases, the system reads the current speed of the high-speed motor in real time, and dynamically calculates the second electrical angular frequency according to the pole pair number of the motor and the base speed signal. The second electrical angular frequency is the synchronous rotation speed of the motor in the physical sense and reflects the real-time state of the stator magnetic field rotation. Based on this, through integral operation, the real-time electrical angular frequency is converted into a phase accumulation quantity, and then the current motor phase angle is obtained at each sampling moment. The d-q coordinate system decomposition is performed on the voltage and current data that have been phase-compensated. In this decomposition process, the electrical signals in the three-phase stationary coordinate system are mapped to two-phase α-β orthogonal components through the Clarke transformation, and with the help of the Park transformation, the α-β components are rotationally transformed into the d-axis and q-axis components that rotate synchronously with the stator magnetic field. This transformation mechanism can simplify the originally complex three-phase alternating signal into two DC quantities with clear physical meaning and strong real-time responsiveness. The d-axis component intuitively represents the magnetic flux excitation component, and the q-axis component directly corresponds to the torque component and the energy conversion dynamics. Combining the above real-time motor phase information, the magnetic flux coefficient matrix is dynamically updated. The magnetic flux coefficient matrix is a core parameter set of the entire orthogonal magnetic flux observation model, and its content reflects the mapping relationship between voltage and current signals, harmonic components, magnetic flux dynamics, and working condition adaptive factors. Due to the continuous changes in the motor operating state, load, temperature, and electromagnetic environment, the magnetic flux coefficient matrix must depend on the current phase information, speed state, and the latest d-q electrical parameters, and dynamically and adaptively adjust its own weights and structure within each control cycle. Specifically, in implementation, based on the current phase, the orthogonal basis functions (such as sin(ωt), cos(ωt) and their higher harmonics) in the auxiliary variable matrix are combined with the latest d-q component weights to solve the real-time numerical values of the current auxiliary variables, and obtain the physical meaning and parameter weights of each component at this moment and under this working condition. After completing the coefficient update, the obtained real-time auxiliary variable values, as well as the current d-axis voltage and current components and q-axis voltage and current components, are input into the orthogonal magnetic flux observation model. This observation model is a set of high-order signal processing algorithms based on frequency-domain orthogonal decomposition, least squares parameter fitting, and physical mechanism constraints. Through matrix operations and linear combinations, the model quickly calculates and outputs the magnitude of the main magnetic flux, harmonic magnetic flux components, and their change rates at the current instant within each sampling period.Based on the real-time observed data of the magnetic flux linkage output, the system accurately depicts the actual dynamics of the magnetic flux linkage of the motor under different operating states, and in real-time feedbacks the physical changes brought about by sudden changes in working conditions, external disturbances or parameter drifts, providing input data for upper-layer algorithm modules such as adaptive regulation, disturbance compensation and control law switching.

[0033] In one embodiment, the adaptive analysis module 13 is specifically configured to: Classify and label the real-time observed data of the magnetic flux linkage, and construct a reference data set for the low-speed region, a reference data set for the medium-speed region, and a reference data set for the high-speed region; Based on the reference data set for the low-speed region, the reference data set for the medium-speed region, and the reference data set for the high-speed region, construct a multi-modal reference model group, which includes a magnetic flux linkage reference model for the low-speed region, a magnetic flux linkage reference model for the medium-speed region, and a magnetic flux linkage reference model for the high-speed region; Perform a fast Fourier transform on the difference between the real-time observed data of the magnetic flux linkage and the output values of the multi-modal reference model group to obtain the magnetic flux linkage observation error spectrum; Separate the magnetic flux linkage observation error spectrum into low-frequency disturbance error and parameter drift error, and perform integral compensation control and parameter adaptive law analysis on the low-frequency disturbance error and parameter drift error to obtain the magnetic flux linkage error compensation amount; Calculate the weight coefficients of the magnetic flux linkage reference model for the low-speed region, the magnetic flux linkage reference model for the medium-speed region, and the magnetic flux linkage reference model for the high-speed region at the current rotational speed and temperature, and perform weighted fusion based on the weight coefficients to obtain the initial magnetic flux linkage reference values for different rotational speed intervals; Superimpose the initial magnetic flux linkage reference values for different rotational speed intervals with the magnetic flux linkage error compensation amount respectively to generate the magnetic flux linkage control reference values for different rotational speed intervals.

[0034] In this embodiment, systematic classification and labeling are performed on the real-time data of the continuously observed magnetic flux linkage. As the motor operates, the system collects a large number of magnetic flux linkage observation values in real time, and each set of observation data corresponds to a specific operating range and operating condition parameters. By dynamically comparing with the motor speed threshold and operating condition parameters, all magnetic flux linkage observation data are automatically divided and classified into three major ranges: low-speed range, medium-speed range, and high-speed range. The labeling helps with subsequent data management and model training, and a highly targeted reference sample set is established for each speed range. The low-speed range reference data set contains all magnetic flux linkage observation values of the motor in states such as low speed, starting, and stable transition. The medium-speed range data set includes typical operating conditions such as normal operation and load changes. The high-speed range reference data set mainly targets the magnetic flux linkage dynamics under extreme or special operating conditions such as field weakening speed increase and high-speed operation. Based on the low-speed range reference data set, medium-speed range reference data set, and high-speed range reference data set, a multi-modal reference model group is constructed. For each speed range, independent low-speed range magnetic flux linkage reference models, medium-speed range magnetic flux linkage reference models, and high-speed range magnetic flux linkage reference models are trained using methods such as machine learning, physical modeling, or system identification. These models respectively learn and fit the dynamic evolution, main frequency changes, harmonic characteristics, and environmental parameter effects of the magnetic flux linkage signals in each range, so that each model can accurately depict the ideal trajectory, dominant frequency, and non-linear response of the magnetic flux linkage under the target speed range. During the actual operation of the motor, the system gradually compares the latest real-time magnetic flux linkage observation data with the output values of the multi-modal reference model group, and calculates the difference between the observation data and the model prediction values at each moment. To analyze and utilize these differences, the fast Fourier transform is performed on the error signal to map the error information in the time domain to the frequency domain, decompose each main frequency, harmonic, and non-stationary component, obtain the magnetic flux linkage observation error spectrum, and reveal the dominant frequency, energy distribution, and periodic disturbances of the current system error. The results of the spectrum analysis show two main components: one is the disturbance component concentrated in the lower frequency band, reflecting low-frequency disturbance errors such as load fluctuations, external mechanical shocks, and power supply disturbances; the other is the parameter drift error distributed in the medium and high frequency bands, manifested as slow drift or parameter changes caused by parameter changes, such as slow-varying errors caused by long-term accumulation of temperature changes, aging, and material property fluctuations. When the system separates the magnetic flux linkage observation error spectrum, methods such as window function segmentation, principal component analysis, and energy spectrum clustering are used to separate the low-frequency disturbance error and the parameter drift error. For the low-frequency disturbance error, an integral compensation control strategy is adopted, and the low-frequency error is integrated and accumulated in the time domain to form a targeted compensation amount, thereby suppressing the steady-state offset and dynamic overshoot caused by external disturbances. For the parameter drift error, the system uses adaptive law analysis, and through means such as gradient descent, parameter regression, and model correction, the parameter set of the multi-modal reference model is corrected in real time to achieve adaptive tracking and structural adjustment of long-term slow-varying disturbances, and then generate a highly robust magnetic flux linkage error compensation amount.According to the working condition parameters such as the rotation speed and temperature collected in real time, the weight coefficients of the flux linkage reference models in the low-speed region, medium-speed region, and high-speed region under the current working conditions are dynamically calculated. Methods such as Gaussian distribution, soft switching, and weighted average are used to set the weight function to ensure smooth transition at the critical points of the speed range and prevent mutations and oscillations during model switching. The adaptive change of the weighting coefficient improves the continuity of the model output and can also highlight the most relevant reference model according to the actual working conditions, making the entire flux linkage control reference value always highly consistent with the actual operating state of the motor. Through weight fusion, the initial flux linkage reference values in different speed ranges are finally obtained. The initial flux linkage reference values in the above different speed ranges are superimposed one by one with the corresponding flux linkage error compensation amounts to generate the final flux linkage control reference value for feedback control.

[0035] In one embodiment, the output module 14 is specifically configured to: Decompose the flux linkage control reference values in different speed ranges to obtain a linear inductance matrix part and a non-linear flux linkage compensation part; Execute current control based on the linear inductance matrix part to obtain a basic control quantity; Input the current current vector and speed data of the high-speed motor into the magnetic-mechanical coupling control algorithm for calculation to obtain a non-linear flux linkage compensation control quantity; Execute control signal synthesis based on the basic control quantity and the non-linear flux linkage compensation control quantity to obtain a synthesized control signal; Identify and compensate for harmonic interference in the synthesized control signal to obtain a compensated control signal, and perform pulse width modulation transformation on the compensated control signal to output a low-noise drive control signal for the motor.

[0036] In this embodiment, the flux control reference values in different speed ranges are decomposed. The flux control reference value integrates the achievements of upper-layer intelligent algorithms such as multi-modal flux observation, working condition self-adaptation, and multi-objective weight adjustment. This reference value not only includes the direct coupling relationship between current and flux under ideal linear working conditions, but also contains the non-linear flux effects, saturation characteristics, and high-order harmonic dynamics of the motor under extreme working conditions such as high speed and field weakening for speed increase. Therefore, the flux control reference value is structurally decomposed from the perspectives of physics and modeling. On the one hand, the main flux part that can directly reflect the linear inductance characteristics is extracted to form the linear inductance matrix part. On the other hand, the remaining components related to non-ideal effects and high-order disturbances are classified as the non-linear flux compensation part. Under the guidance of the linear inductance matrix part, the system enters the basic link of current control. Based on the d-q axis space vector theory and the principle of stator flux orientation control, the controller uses the linear inductance matrix and the target flux reference to calculate the d-axis and q-axis current commands in the ideal state in real time. Combining with the currently observed current components, the basic control quantity is generated through proportional-integral or higher-level linear adjustment algorithms. The physical essence of the basic control quantity is to ensure that the stator current of the motor can accurately match the flux target at any time, realizing stable excitation, accurate speed regulation, and low torque ripple, and laying the foundation for the baseline performance of the entire system. This part of the control loop has the advantages of simple structure, fast dynamic response, and easy engineering implementation. However, it is difficult for a single linear inductance model to cover various non-linear phenomena exhibited by high-speed motors during actual operation. Especially under extreme working conditions such as high speed, field weakening for speed increase, and load mutation, the core saturation, flux distortion, cross-coupling, and eddy current effects of the motor will significantly affect the flux dynamic behavior. Therefore, a high-order non-linear flux compensation mechanism is designed. The output module feeds key parameters such as the current vector and real-time speed of the motor as input quantities into the magnetic-mechanical coupling control algorithm. Based on advanced methods such as neural networks, fuzzy logic, and adaptive non-linear modeling, this algorithm real-time identifies and compensates for non-linear flux disturbances and complex dynamic relationships. Inside the algorithm, physical quantities such as current, flux, and speed in historical and current states are used to dynamically estimate the flux error caused by factors such as magnetic saturation, iron loss, and non-ideal electromagnetic characteristics, and output it as the non-linear flux compensation control quantity. Through the adaptive compensation mechanism, even when the motor is in the field weakening region, high-frequency disturbance, or extreme load switching environment, it can restore the flux target value to the greatest extent, reduce the flux fluctuation, and effectively suppress harmonic excitation, resonance, and energy loss. The basic control quantity and the non-linear flux compensation control quantity are fused through an intelligent synthesis strategy. The synthesis mechanism adopts various methods such as weight distribution, parallel superposition, and segmented selection. The core of its design is to ensure the seamless switching and collaborative gain of the high response speed of linear regulation and the high robustness of non-linear compensation under different operating conditions. The synthesized control signal retains the forced constraint of the basic control on the main component of the flux and introduces the adaptive correction ability of non-linear compensation for high-order errors.Identify and compensate for harmonic interference in the synthesized control signal. Through algorithms such as frequency-domain analysis, band-stop notch filtering, and harmonic reconstruction, actively identify the high-frequency disturbance components in the synthesized control signal. For the main harmonic components, adopt control means such as notch filtering, adaptive band-stop, and inverse harmonic injection to implement active compensation, effectively suppress the high-frequency interference within the control loop, and form a multi-link joint suppression of harmonic noise in the motor body and drive link, thereby minimizing the overall system noise level and the amplitude of magnetic flux ripple to the greatest extent. After completing the harmonic interference compensation, transmit the compensated control signal to the pulse-width modulation conversion unit. Pulse-width modulation discretizes the continuous control signal into drive pulses executable by the inverter, and through the adaptive adjustment of the carrier modulation ratio, carrier synchronization, and switching strategy, realizes the high-frequency smoothing and energy optimization of the signal. During the pulse-width modulation conversion process, analyze the spectral characteristics of the compensated control signal, dynamically adjust the modulation parameters, and output the low-noise drive control signal for the motor.

[0037] In one embodiment, the low-noise magnetic flux adaptive feedback control system for a high-speed motor further includes: A hierarchical control module, used to distribute the low-noise drive control signal of the motor to the basic control layer, the adaptive estimation layer, and the intelligent optimization layer to obtain hierarchical control instructions; based on the noise characteristic data of the high-speed motor, set different execution priorities and control bandwidths for the hierarchical control instructions to obtain a priority control strategy; perform basic motor current regulation according to the control instructions of the basic control layer in the hierarchical control instructions to obtain a basic current regulation amount; perform dynamic compensation on the basic current regulation amount based on the control instructions of the adaptive estimation layer in the hierarchical control instructions to obtain a compensated current control amount; use the control instructions of the intelligent optimization layer in the hierarchical control instructions to perform smooth transition control on the sudden operating conditions of the high-speed motor to generate a smooth transition control signal; perform coordinated synthesis and dual-sampling rate processing on the compensated current control amount and the smooth transition control signal, and output a full-speed-range low-noise current control instruction.

[0038] In this embodiment, the hierarchical control module receives the motor low-noise drive control signal obtained after harmonic suppression and pulse width modulation from the output module, and decomposes this signal into hierarchical control instructions at three levels according to the working condition perception, process requirements, and real-time operating state. The basic control layer focuses on the basic regulation of the motor current. Its goal is to quickly respond to the target current instruction, achieve the basic closed-loop regulation of flux linkage, torque, and speed, and ensure that the system can operate stably at the optimal response speed and minimum steady-state error under most steady-state and slowly changing working conditions. The basic control layer compares the actual current with the target current according to the current target in the hierarchical control instruction and quickly corrects it through high-performance current loop algorithms such as proportional-integral, feedforward compensation, and space vector pulse width modulation, and outputs the basic current regulation amount. The hierarchical control module introduces an adaptive estimation layer and an intelligent optimization layer to perform real-time compensation and global optimization on the basic current regulation amount. In the adaptive estimation layer, the hierarchical control instruction will guide the system to continuously and dynamically correct the current deviation caused by model parameter drift, load change, environmental noise, etc. through algorithms such as model reference adaptive, adaptive observer, and robust feedback regulation based on multi-source data such as flux linkage observation, parameter identification, and noise spectrum analysis. The essence of the adaptive estimation layer is to highly couple the theoretical model with the actual operation in a way that combines data-driven and model-driven, and ensure the system's sensitive perception and immediate response to complex disturbances and dynamic changes throughout the entire life cycle, and output the compensated current control amount. At the same time, the intelligent optimization layer serves as the top-level scheduling center to identify and handle the global response optimization and smooth control transition generated by high-speed motors during sudden changes in working conditions such as rapid acceleration, rapid deceleration, and sudden load changes. The intelligent optimization layer combines historical operation data, real-time noise characteristics, load trends, and predictive analysis, and comprehensively considers multi-objective requirements such as flux linkage stability, noise suppression, and energy efficiency maximization through multi-objective genetic algorithms, fuzzy inference, neural networks, or multi-variable optimization methods. When it is detected that the system is about to enter or has entered the sudden change area of the working condition, the intelligent optimization layer generates a smooth transition control signal in advance to perform secondary correction and buffering on the basic regulation amount and the adaptive compensation amount, and eliminate the current impact, large flux linkage fluctuations, and high-frequency noise excitation caused by the sharp change. The smooth transition signal is a means of global coordination and dynamic adjustment, which can dynamically balance secondary objectives (such as energy efficiency or response speed) on the basis of ensuring the main objectives (such as flux linkage steady state and noise suppression), and achieve the smooth and optimal transfer of the control effect. The hierarchical control module assigns appropriate execution priorities and control bandwidths to each layer based on the noise characteristic data of the high-speed motor and the current working condition requirements. The basic control layer has the highest priority and the largest bandwidth to handle the high-speed changes and active dynamic responses of the current loop; the adaptive estimation layer focuses on disturbance compensation and model adaptation; the intelligent optimization layer aims at global optimality and smooth working condition switching, and its bandwidth and real-time performance are appropriately sacrificed for the overall strategy optimization.The introduction of the priority control strategy effectively prevents the algorithms at each level from interfering with each other in a conflict state, and ensures that the control focus can automatically shift to the most critical tasks under special working conditions, maximizing the overall robustness and adaptability of the system. The compensated current control quantity and the smooth transition control signal are adjusted in weight, resolved in conflict, and fused in signal through the coordinated synthesis mechanism of the hierarchical control module, ensuring that the total control command output has multiple characteristics such as fast response, robust compensation, and smooth transition. To improve the full-speed performance and system stability, the hierarchical control module introduces a dual sampling rate processing mechanism: for the basic control layer and the adaptive estimation layer, a low sampling rate is used to reduce the computational pressure and enhance the anti-interference ability of the signal; while the intelligent optimization layer and the working condition smooth control closely follow the working condition changes at a high sampling rate, improving the fineness and flexibility of the mutation response. Output a full-speed low-noise current control command.

[0039] In the embodiments of the present invention, through the frequency adaptive auxiliary variable and the orthogonal flux linkage observer, high-frequency interference and noise are effectively filtered, the propagation of electromagnetic noise is suppressed from the source, the electromagnetic noise and mechanical vibration during the operation of the high-speed motor are greatly reduced, and the operation comfort and stability of the equipment are improved. The multi-modal model reference adaptive system is adopted to divide the motor working range into multiple modes and achieve smooth transition. Combining the spectrum analysis mechanism of the flux linkage observation error, the errors caused by load disturbance and parameter drift are distinguished, and high-precision flux linkage control performance is maintained within the full speed range. The nonlinear flux linkage model is decomposed through a special coordinate transformation, and combined with the adaptive high-order neural network structure, the complex flux linkage nonlinear relationship of the high-speed motor in the field weakening region is accurately modeled, effectively dealing with the problem of flux path distortion under high-speed operation. A smooth transition mechanism is designed to ensure smooth transition of the control output during large-scale speed changes, avoid control oscillation, significantly improve the stability and comfort of the system during rapid speed changes, and reduce speed fluctuations and abnormal noises. The real-time self-calibration mechanism combined with the working condition identifier can maintain the best performance during long-term operation and various working conditions, automatically select the optimization strategy suitable for the current working condition, and maintain consistent control performance under extreme temperature and complex load conditions. The dual sampling rate control strategy and the hierarchical control architecture are adopted. The low sampling rate subsystem is used for the basic control layer and the adaptive estimation layer, and the high sampling rate subsystem is used for the intelligent optimization layer and the spectrum analysis, optimizing the system resource allocation, reducing the control delay, and improving the overall control efficiency. A real-time self-calibration mechanism is designed to complete parameter update during normal operation by injecting detection signals of specific frequencies and detecting the response characteristics, improving the continuous availability of the system and avoiding the downtime and efficiency loss caused by traditional offline calibration methods.

[0040] It should be noted that in this text, the term "including", "comprising", or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, system, article, or system that includes a series of elements not only includes those elements but also includes other elements not explicitly listed, or further includes elements inherent to such process, system, article, or system. Without further limitation, an element defined by the statement "including one..." does not exclude the presence of additional identical elements in the process, system, article, or system that includes such element.

[0041] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A low-noise magnetic flux adaptive feedback control system for a high-speed motor, characterized in that, Including: A collection module, which is used to collect current, voltage and vibration data of a high-speed motor at various operating speeds, and obtain the operating parameters and noise characteristic data of the high-speed motor; A transformation processing module, which is used to perform frequency-domain orthogonal transformation processing on the operating parameters of the high-speed motor and the noise characteristic data to obtain real-time magnetic flux observation data; An adaptive analysis module, which is used to perform multi-modal condition adaptive analysis according to the real-time magnetic flux observation data and generate magnetic flux control reference values for different speed intervals; An output module, which is used to input the magnetic flux control reference values for different speed intervals into a magnetic-mechanical coupling control algorithm, calculate and output a low-noise drive control signal for the motor.

2. The low-noise flux adaptive feedback control system of the high-speed motor according to claim 1, characterized in that Specifically, the collection module is used for: Obtaining the speed data of the high-speed motor, and dividing the working range of the high-speed motor into a low-speed area, a medium-speed area and a high-speed area based on the speed data; Injecting frequency detection signals into the low-speed area, the medium-speed area and the high-speed area respectively, and collecting the three-phase voltage signals and three-phase current signals of the high-speed motor in the low-speed area, the medium-speed area and the high-speed area; Performing Clarke transformation and Park transformation processing on the three-phase voltage signals to obtain d-q axis voltage data, and performing Clarke transformation and Park transformation processing on the three-phase current signals to obtain d-q axis current data; Collecting the mechanical vibration signals and electromagnetic noise signals generated during the operation of the high-speed motor in the low-speed area, the medium-speed area and the high-speed area, and performing fast Fourier transform on the mechanical vibration signals and the electromagnetic noise signals respectively to obtain noise characteristic data; Integrating the speed data, the d-q axis voltage data and the d-q axis current data to obtain the operating parameters of the high-speed motor.

3. The low-noise magnetic flux adaptive feedback control system of a high-speed motor according to claim 2, characterized in that The transformation processing module further includes: A construction unit, which is used to construct a frequency-domain orthogonal basis matrix including sine functions and cosine functions according to the speed data in the operating parameters of the high-speed motor; An orthogonal transformation unit, which is used to perform orthogonal transformation on the d-q axis voltage data and the d-q axis current data by using the frequency-domain orthogonal basis matrix to obtain frequency-domain transformed electrical parameter data; A spectrum analysis unit, which is used to perform spectrum analysis on the noise characteristic data to obtain high-order harmonic components and slot harmonic components, and construct a harmonic interference model according to the high-order harmonic components and the slot harmonic components; A filtering processing unit, which is used to perform harmonic noise identification and filtering processing on the frequency-domain transformed electrical parameter data based on the harmonic interference model to obtain filtered electrical parameters; A magnetic flux observation unit, which is used to calculate a magnetic flux coefficient matrix by using the filtered electrical parameters and construct an orthogonal magnetic flux observation model according to the magnetic flux coefficient matrix; A magnetic flux online calculation unit, which is used to obtain the real-time voltage data and real-time current data of the high-speed motor, input them into the orthogonal magnetic flux observation model for magnetic flux online calculation, and output real-time magnetic flux observation data.

4. The low-noise magnetic flux adaptive feedback control system of a high-speed motor according to claim 3, characterized in that, Specifically, the magnetic flux observation unit is used for: Converting the speed data in the operating parameters of the high-speed motor into a first electrical angular frequency, and constructing a frequency adaptive auxiliary variable matrix based on the first electrical angular frequency; Perform Fourier decomposition on the filtered electrical parameters to obtain a relationship curve between the flux linkage component and the frequency, and determine the harmonic order based on the relationship curve between the flux linkage component and the frequency; Construct an orthogonal basis function group according to the harmonic order and the frequency adaptive auxiliary variable matrix, and expand the flux linkage signal using the orthogonal basis function group to obtain a flux linkage orthogonal decomposition structure; Perform matrix multiplication on the filtered electrical parameters and the frequency adaptive auxiliary variable matrix to obtain an auxiliary variable covariance matrix; Add a regularization term to the auxiliary variable covariance matrix to obtain an enhanced covariance matrix, and solve the regularized least squares equation using the enhanced covariance matrix to obtain a flux linkage coefficient matrix; Construct a flux linkage observation equation based on the flux linkage coefficient matrix and the flux linkage orthogonal decomposition structure, and replace the time variable in the flux linkage observation equation with the phase angle calculated online to obtain an orthogonal flux linkage observation model.

5. The low-noise flux adaptive feedback control system of a high-speed motor according to claim 4, wherein The flux linkage online calculation unit is specifically used for: Collect the real-time voltage data and real-time current data of the high-speed motor, and perform phase compensation on the real-time voltage data and the real-time current data to obtain phase-compensated voltage and current data; Calculate the second electrical angular frequency according to the current speed of the high-speed motor, and calculate the phase angle at the current moment based on the second electrical angular frequency to obtain the real-time phase information of the motor; Decompose the phase-compensated voltage and current data in the d-q coordinate system to obtain d-axis voltage and current components and q-axis voltage and current components; Update the flux linkage coefficient matrix according to the real-time phase information of the motor to obtain a real-time auxiliary variable value; Substitute the real-time auxiliary variable value, the d-axis voltage and current components, and the q-axis voltage and current components into the orthogonal flux linkage observation model for flux linkage calculation to obtain real-time flux linkage observation data.

6. The low-noise flux self-adaptive feedback control system of a high-speed motor according to claim 1, wherein The adaptive analysis module is specifically used for: Classify and label the real-time flux linkage observation data to construct a reference data set for the low-speed region, a reference data set for the medium-speed region, and a reference data set for the high-speed region; Construct a multi-modal reference model group based on the reference data set for the low-speed region, the reference data set for the medium-speed region, and the reference data set for the high-speed region. The multi-modal reference model group includes a low-speed region flux linkage reference model, a medium-speed region flux linkage reference model, and a high-speed region flux linkage reference model; Perform a fast Fourier transform on the difference between the real-time flux linkage observation data and the output values of the multi-modal reference model group to obtain a flux linkage observation error spectrum; Separate the flux linkage observation error spectrum into a low-frequency disturbance error and a parameter drift error, and perform integral compensation control and parameter adaptive law analysis on the low-frequency disturbance error and the parameter drift error to obtain a flux linkage error compensation amount; Calculate the weight coefficients of the low-speed region flux linkage reference model, the medium-speed region flux linkage reference model, and the high-speed region flux linkage reference model at the current speed and temperature, and perform weighted fusion based on the weight coefficients to obtain initial flux linkage reference values for different speed intervals; Superimpose the initial flux linkage reference values for different speed intervals and the flux linkage error compensation amount respectively to generate flux linkage control reference values for different speed intervals.

7. The low-noise magnetic flux adaptive feedback control system of a high-speed motor according to claim 1, characterized in that The output module is specifically configured to: Decompose the flux control reference values in different speed ranges to obtain a linear inductance matrix part and a non-linear flux compensation part; Execute current control based on the linear inductance matrix part to obtain a basic control quantity; Input the current current vector and speed data of the high-speed motor into the magnetic-mechanical coupling control algorithm for calculation to obtain a non-linear flux compensation control quantity; Execute control signal synthesis based on the basic control quantity and the non-linear flux compensation control quantity to obtain a synthesized control signal; Identify and compensate for harmonic interference in the synthesized control signal to obtain a compensated control signal, and perform pulse width modulation transformation on the compensated control signal to output a low-noise drive control signal for the motor.

8. The low-noise flux adaptive feedback control system of the high-speed motor according to claim 1, characterized in that, The low-noise flux adaptive feedback control system of the high-speed motor further includes: A hierarchical control module, which is used to distribute the low-noise drive control signal of the motor to the basic control layer, the adaptive estimation layer, and the intelligent optimization layer to obtain hierarchical control instructions; based on the noise characteristic data of the high-speed motor, set different execution priorities and control bandwidths for the hierarchical control instructions to obtain a priority control strategy; perform basic motor current regulation according to the control instructions of the basic control layer in the hierarchical control instructions to obtain a basic current regulation quantity; perform dynamic compensation on the basic current regulation quantity based on the control instructions of the adaptive estimation layer in the hierarchical control instructions to obtain a compensated current control quantity; use the control instructions of the intelligent optimization layer in the hierarchical control instructions to perform smooth transition control on the sudden operating conditions of the high-speed motor to generate a smooth transition control signal; perform coordinated synthesis and dual-sampling rate processing on the compensated current control quantity and the smooth transition control signal to output a full-speed range low-noise current control instruction.

Images