A vector control method and system based on multi-dimensional space

Abstract

The embodiment of the application provides a kind of vector control method and system based on multidimensional space, it is related to the technical field of motor technology.The method comprises: obtaining the real-time current error signal of the current loop of motor controller;Based on the real-time current error signal, selective harmonic analysis is performed on the preset key frequency point to construct the multi-dimensional vibration vector representing the vibration state of the system, wherein the key frequency point covers the audible frequency range of human ear;The multi-dimensional vibration vector is first processed to calculate the perceived vibration energy value;According to the deviation between the perceived vibration energy value and the preset target energy value, generate proportional-integral parameter adjustment amount;Based on the proportional-integral parameter adjustment amount, dynamically update the proportional-integral controller parameters of the current loop.Through the application, the problem of low motor control sensing accuracy is solved, and the sensing accuracy is improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of motor control, and more specifically, to a vector control method and system based on multi-dimensional space. Background Technology

[0002] In modern high-performance servo drive and frequency converter control systems, the current loop, as the innermost control element, directly determines the system's dynamic response speed, stability, and anti-interference capability. Proportional-integral (PI) controllers are widely used in current loop control due to their simple structure, ease of implementation, and robustness.

[0003] Traditional PI controller parameters (proportional gain Kp and integral gain Ki) are typically set to fixed values ​​or tuned in segments only at a limited number of operating points. This static parameter setting method is insufficient to effectively handle the complex disturbances encountered by motor systems in actual operation, such as high-frequency periodic disturbances caused by load changes, mechanical structure resonance, inverter nonlinearity, and electromagnetic noise. In particular, when current fluctuations occur in the audible frequency range of 1-20kHz during system operation, they can easily excite electromagnetic vibrations in the motor stator, rotor, or the entire mechanical structure, generating harsh electromagnetic howling noise. This problem is especially prominent in fields with stringent requirements for noise, vibration, and harshness (NVH), such as household appliances (e.g., inverter air conditioners, washing machines), electric vehicle drive systems, and precision instruments.

[0004] To address this issue, existing technologies have proposed intelligent adjustment methods such as fuzzy control PI and neural network adaptive PI. However, these methods generally suffer from low sensing accuracy, leading to increased hardware costs and difficulties in engineering applications. Summary of the Invention

[0005] This invention provides a vector control method and system based on multi-dimensional space, which at least solves the problem of low perception accuracy in related technologies.

[0006] According to an embodiment of the present invention, a vector control method based on multi-dimensional space is provided, comprising:

[0007] Obtain the real-time current error signal of the current loop of the motor controller;

[0008] Based on the real-time current error signal, selective harmonic analysis is performed on preset key frequency points to construct a multidimensional vibration vector characterizing the vibration state of the system, wherein the key frequency points cover the audible frequency band.

[0009] The multidimensional vibration vector is first processed to calculate the perceived vibration energy value;

[0010] Based on the deviation between the perceived vibration energy value and the preset target energy value, a proportional-integral parameter adjustment amount is generated;

[0011] The proportional-integral controller parameters of the current loop are dynamically updated based on the proportional-integral parameter adjustment.

[0012] In one exemplary embodiment, obtaining the real-time current error signal of the motor controller current loop includes:

[0013] Real-time acquisition of three-phase current signals from the motor;

[0014] A coordinate transformation is performed on the three-phase current signal to obtain the d-axis current component and the q-axis current component in the rotating coordinate system;

[0015] The d-axis current component and q-axis current component are compared with the d-axis command current and q-axis command current, respectively, to obtain the d-axis current error and q-axis current error, and the q-axis current error is used as the real-time current error signal.

[0016] In one exemplary embodiment, the first processing of the multidimensional vibration vector includes:

[0017] Each element of the multidimensional vibration vector is weighted and combined with the weighting coefficient of the corresponding frequency point to obtain the perceived vibration energy value.

[0018] In one exemplary embodiment, after dynamically updating the proportional-integral controller parameters of the current loop, the method further includes:

[0019] When the preset active detection conditions are met, a broadband detection signal is injected into the current loop;

[0020] The system response signal generated by the broadband detection signal is collected and analyzed to identify the system gain characteristics of the current loop at key frequency points, so as to obtain the first spectrum.

[0021] Multi-objective optimization is performed based on the perceived vibration energy value and the first spectrum to update the proportional-integral controller parameters.

[0022] In an exemplary embodiment, performing multi-objective optimization based on the auditory vibration energy value and the first spectrum includes:

[0023] The perceived vibration energy value is fused with the maximum gain peak value contained in the first spectrum to obtain a cost function;

[0024] Adjust the proportional-integral controller parameters to minimize the cost function.

[0025] According to another embodiment of the present invention, a vector control system based on multi-dimensional space is provided, comprising:

[0026] The current error acquisition module is used to acquire the real-time current error signal of the current loop of the motor controller;

[0027] The vibration vector construction module is used to perform selective harmonic analysis on preset key frequency points based on the real-time current error signal in order to construct a multidimensional vibration vector characterizing the vibration state of the system, wherein the key frequency points cover the audible frequency band.

[0028] The auditory energy assessment module is used to perform a first processing on the multidimensional vibration vector to calculate the auditory vibration energy value;

[0029] The parameter adjustment module is used to generate a proportional-integral parameter adjustment amount based on the deviation between the perceived vibration energy value and the preset target energy value.

[0030] The parameter update module is used to dynamically update the proportional-integral controller parameters of the current loop based on the proportional-integral parameter adjustment amount.

[0031] In one exemplary embodiment, obtaining the real-time current error signal of the motor controller current loop includes:

[0032] Real-time acquisition of three-phase current signals from the motor;

[0033] A coordinate transformation is performed on the three-phase current signal to obtain the d-axis current component and the q-axis current component in the rotating coordinate system;

[0034] The d-axis current component and q-axis current component are compared with the d-axis command current and q-axis command current, respectively, to obtain the d-axis current error and q-axis current error, and the q-axis current error is used as the real-time current error signal.

[0035] In one exemplary embodiment, the first processing of the multidimensional vibration vector includes:

[0036] Each element of the multidimensional vibration vector is weighted and combined with the weighting coefficient of the corresponding frequency point to obtain the perceived vibration energy value.

[0037] According to yet another embodiment of the present invention, a computer-readable storage medium is also provided, wherein a computer program is stored therein, wherein the computer program is configured to perform the steps in any of the above method embodiments when executed.

[0038] According to yet another embodiment of the present invention, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0039] This invention constructs a multidimensional vibration vector, thereby adjusting a single scalar current error signal into a state vector in a specific frequency space. It focuses on several key frequencies that have the greatest impact on human auditory experience, achieving targeted analysis of harmful harmonics. This improves the perception dimension and identification accuracy of the control system for vibration noise sources, thus solving the problem of low perception accuracy and achieving the effect of improving perception accuracy. Attached Figure Description

[0040] Figure 1 This is a flowchart of a vector control method based on multi-dimensional space according to an embodiment of the present invention;

[0041] Figure 2 This is a structural block diagram of a vector control system based on a multi-dimensional space according to an embodiment of the present invention;

[0042] Figure 3 This is a resonance vulnerability map according to an embodiment of the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of this application will be described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0044] In the following description, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0045] Furthermore, in this application, directional terms such as "upper," "lower," "left," and "right" may be defined relative to the orientation of the components shown in the accompanying drawings. It should be understood that these directional terms can be relative concepts, used for relative description and clarification, and may change accordingly depending on the orientation of the components in the accompanying drawings.

[0046] In this application, unless otherwise expressly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral part; it can be a direct connection or an indirect connection through an intermediate medium. Furthermore, the term "coupled" can refer to an electrical connection that enables signal transmission.

[0047] As used herein, “about,” “approximately,” or “approximately” includes the stated value and the average value within an acceptable range of deviation from the given value, wherein the acceptable range of deviation is determined by a person skilled in the art taking into account the measurement under discussion and the error associated with the measurement of the given quantity (i.e., the limitations of the measurement system).

[0048] Example 1

[0049] This embodiment provides a vector control method based on multi-dimensional space. This method achieves online adaptive adjustment of PI controller parameters by performing in-depth analysis of the error signal of the current loop, thereby actively suppressing vibrations at specific frequencies and eliminating or significantly reducing electromagnetic noise.

[0050] Reference Figure 1 This method can be broken down into the following steps:

[0051] S100: Obtain the real-time current error signal of the motor controller current loop.

[0052] In this embodiment, the system is applied to a permanent magnet synchronous motor (PMSM) servo system employing a field-oriented control (FOC) strategy. In this system, the current loop is the fastest-responding inner loop, enabling the actual stator current of the motor to accurately follow the command current. The real-time current error signal is used to reflect the current tracking performance, ensuring that the system can capture in real time the non-ideal current components generated by various disturbances (such as load fluctuations, mechanical resonance, inverter dead-zone effects, etc.), which are the root causes of vibration and noise.

[0053] Specifically, firstly, the system's current sampling unit (e.g., a Hall sensor or sampling resistor connected in series with the three-phase windings of the motor) uses a sampling frequency much higher than the motor's electrical frequency. For three-phase current High-speed data acquisition is performed. This sampling frequency... The sampling frequency should be at least twice the highest frequency required for analysis; considering that the upper limit of the audible frequency range for the human ear is 20kHz, the sampling frequency should be at least twice the highest frequency required for accurate analysis of harmonics within this range. It is typically set to 40kHz or higher. For example, in a high-performance servo drive, the sampling frequency can be set. kHz (unit: Hertz) to ensure sufficient analysis margin for frequency components up to 25 kHz.

[0054] The collected three-phase alternating current Being in a stationary ABC coordinate system makes decoupling control inconvenient; therefore, coordinate transformation is required. In this case, the Clarke transformation can be used to transform the three-phase current into a two-phase stationary coordinate system. Current components in coordinate system and Subsequently, the real-time rotor electrical angle provided by the motor rotor position sensor (such as an encoder) is utilized. Through the Park transformation, and Transforming to a two-phase rotating coordinate system (dq coordinate system) that rotates synchronously with the rotor magnetic field, the d-axis current component is obtained. and q-axis current component In the FOC control of a PMSM, the d-axis current is typically used to control the magnetic flux, while the q-axis current is proportional to the electromagnetic torque output by the motor.

[0055] Finally, the error calculation unit will feed back the d-axis and q-axis current components in real time. d-axis command current from the upper velocity loop or position loop and q-axis command current By comparing the results, the real-time current error signal is calculated; that is, the d-axis current error. and q-axis current error Since the torque pulsation and related mechanical vibration of the motor are mainly caused by the fluctuation of the q-axis current, in this embodiment, the q-axis current error is selected. As a real-time current error signal.

[0056] For example, suppose that at a certain moment, the controller gives the q-axis command current as follows: A (unit: ampere), while due to load disturbance, the actual q-axis current obtained after system sampling and transformation is: A (unit: ampere), then the instantaneous current error at this time A (unit: ampere); the controller will then continuously execute this process to generate a discrete-time series representing error fluctuations. And so on.

[0057] S200: Based on the real-time current error signal, selective harmonic analysis is performed on multiple key frequency points within the preset audible frequency band to construct a multidimensional vibration vector characterizing the vibration state of the system.

[0058] In this embodiment, firstly, a set of key frequencies needs to be predefined. The M frequency points in this set are selected based on the human hearing sensitivity curve (such as the equal loudness curve) and the common noise frequency distribution characteristics of motor systems. They are usually concentrated in the range of 1kHz to 20kHz, which includes the 2kHz to 5kHz region where the human ear is most sensitive, as well as frequency points that are prone to generating strong harmonics, such as motor cogging effect and inverter switching frequency sideband. The number of M can be balanced according to the processor performance and control accuracy requirements, and is usually between 8 and 16.

[0059] The key frequency set can be defined as follows: In a semi-anechoic chamber environment, the drive motor operates at different speeds under multiple typical operating conditions (e.g., no-load, 50% rated load, 100% rated load), while the noise signal spectrum generated by the motor is recorded using a high-precision microphone and a spectrum analyzer. By analyzing these spectrum diagrams, the peak noise frequencies that persist and have high amplitudes under multiple operating conditions are identified. These frequency points, experimentally verified as the main noise sources, are used as the key frequency set. The key frequency points. For example, after the above calibration process, a set of key frequencies for a specific model of compressor motor can be set as follows: Hz (unit: Hertz), with a total of M=10 key frequency points. The dimension M of this set determines the dimension of the multidimensional vibration vector.

[0060] Next, this embodiment uses the Gossel algorithm to perform selective harmonic analysis. The Gossel algorithm is an extremely efficient digital signal processing technique specifically designed to calculate the energy (or DFT component) of a discrete signal at a single preset frequency point, with a computational complexity of linear order. Far lower than FFT Specifically, the system processes the acquired q-axis current error signal sequence within a data window of fixed length N. M Gossel algorithm filters are applied in parallel, each filter being applied to... Key frequencies Configure it.

[0061] In practical engineering implementation, to ensure the robustness and accuracy of the analysis, the following technical details need to be considered. First, to suppress spectral leakage (i.e., energy "leaking" from one frequency to adjacent frequency points, causing measurement inaccuracies), the signal sequence within the data window must be processed before feeding the data into the Gossel algorithm. First, a window function (such as a Hamming window or a Blackman window) is used to smooth signal abrupt changes at both ends of the data window, resulting in a more concentrated main lobe and faster side lobe attenuation in the frequency domain. Second, to balance real-time performance and frequency resolution, an overlapping sliding mechanism is used for data window processing (the overlap rate of the data window can be set to 50%). This means that whenever N / 2 new sampling points are acquired, a new data window (containing N / 2 old data points and N / 2 new data points) is formed for analysis. This approach ensures the update frequency of the analysis results while utilizing a longer data sequence to improve frequency resolution. Third, when implementing on a fixed-point DSP or a resource-constrained MCU, the intermediate state variables in the Gossel algorithm's IIR filter need to be appropriately calibrated or shifted to prevent numerical overflow during iterative calculations and ensure computational stability.

[0062] Specifically, for each key frequency The system first determines the sampling frequency. Calculate a coefficient in advance Then, within a data window of length N that has been processed by a window function, the error signal is filtered by a second-order IIR filter. ( After iterative calculations are performed, and the results are obtained through combined calculations, the frequency can be determined. Energy amplitude at .

[0063] After processing by M parallel Gosser algorithms, the system will obtain M energy amplitude values. The M amplitude values ​​correspond to the vibration intensity of the system at M key frequency points. This ordered set of amplitude values ​​is set as an M-dimensional vector, i.e., a multidimensional vibration vector. This vector In the M-dimensional frequency characteristic space, its direction and modulus together describe the distribution and overall intensity of the current system's vibrational energy. For example, if the vector is in the third dimension (corresponding to frequency)... ) of the components A particularly large value indicates that the system is... Significant harmonic vibrations exist at certain frequencies.

[0064] For example, assuming the data window size sampling frequency Hz, key frequency set selects two points Assume the error signal is collected within an N-point window. After applying the Hamming window, and through analysis by the Gossel algorithm module, the output is: The energy amplitude at Hz is A (unit: ampere), in The energy amplitude at Hz is A (unit: ampere); then, the constructed two-dimensional vibration vector is: This vector visually indicates that the vibration intensity of the current system at 10500 Hz is greater than that at 4000 Hz.

[0065] In addition, the vibration vector can also be constructed using a set of parallel second-order infinite impulse response (IIR) bandpass filters. Specifically, for each key frequency... Set up an IIR bandpass filter with its center frequency precisely set to... Its bandwidth is selectively set according to the required frequency. For example, it can be set to 5% of the center frequency, i.e., the bandwidth is... Real-time current error signal The signal is simultaneously fed into M parallel bandpass filters, and the output signal of each filter contains only the energy of the original signal within its corresponding key frequency band. Then, the root mean square (RMS) value of the energy of each filter's output signal is calculated within a sliding time window. Thus, at any given time, M RMS values ​​can be obtained.

[0066] This set of RMS values ​​also constitutes an M-dimensional vibration vector. It can also characterize the distribution of system vibration energy in different key frequency bands, and this method is suitable for situations where computing resources are extremely limited.

[0067] S300: It uses a preset auditory weighting model to process multidimensional vibration vectors to calculate the weighted auditory vibration energy value.

[0068] Because the human ear's sensitivity to different frequencies varies greatly, it is necessary to perform auditory correction on the vibration vector of purely physical quantities to make the calculation results more consistent with human subjective perception, thereby allowing parameter adjustments to be more targeted. For example, noise generated by a current fluctuation with an amplitude of 0.1A at 10kHz may sound more jarring than noise generated by a fluctuation with an amplitude of 0.2A at 1kHz. Therefore, the purpose of this step is to introduce a psychoacoustic model.

[0069] This embodiment uses a hearing weighting model based on a Gaussian kernel function. The model takes the form of ,in, It is the center frequency to which the human ear is most sensitive; It is the standard deviation, which determines the range of the sensitive frequency band; these two parameters and The selection of the center frequency needs to be adjusted according to the specific application of the product and the target user group; for example, for medical equipment or high-end home appliances that need to work in a quiet environment, people are most sensitive to and irritated by noise in the 2kHz-5kHz range. In this case, the center frequency can be adjusted accordingly. Set to 3500Hz (unit: Hertz), and set the standard deviation. Set it to a smaller value, such as 1500Hz (unit: Hertz), to form a narrow and sharp weighted curve, focusing on suppressing noise in this frequency band; however, for applications such as industrial robots or electric vehicles, high-frequency howling (such as above 8kHz) may be the main source of user complaints, in which case... Accordingly, the frequency was increased to 10000 Hz (in Hertz), and a larger standard deviation was set. For example, 4500Hz (unit: Hertz) to cover a wider range of high frequencies, and so on.

[0070] Prior to this, the system would calculate the set of key frequencies based on this Gaussian kernel function. Each frequency auditory weighting coefficient This set of weighting coefficients This constitutes a sound perception weight vector.

[0071] The multidimensional vibration vector will then be constructed. With auditory weight vector Perform a weighted summation (or dot product) to obtain a scalarized weighted perceived vibration energy value (i.e., perceived vibration energy value). :

[0072]

[0073] Should The value is the multidimensional vibration vector. In a vector of auditory weights The projection onto the defined "auditory sensitivity axis" is an index of the perceptible noise intensity after filtering by the characteristics of human hearing; when A large value indicates that the system has strong vibrations in the frequency band that the human ear is sensitive to, and these vibrations need to be suppressed.

[0074] For example, the vibration vector is The corresponding frequency is Hz, Hz. Assuming the parameters of the hearing-weighted model are set to... Hz, If the frequency is Hz, then the weighting coefficients corresponding to the two frequency points can be calculated:

[0075]

[0076]

[0077] It's easy to understand that since 10500Hz is exactly the set most sensitive frequency, its weight is 1.0, while 4000Hz is far from the center frequency, its weight is reduced to 0.35.

[0078] Next, calculate the weighted auditory vibration energy value: The result (0.605A) is smaller than the sum of the amplitudes (0.3 + 0.5 = 0.8A), but the contribution of the 10500Hz component is much greater than that of the 4000Hz component, reflecting the difference in the sensitivity of the human ear to noise at these two frequencies.

[0079] S400: Generates a proportional-integral (PI) parameter adjustment amount based on the deviation between the weighted auditory vibration energy value and the preset target energy value;

[0080] S500: Based on the adjustment amount of the PI parameter, dynamically update the proportional-integral controller parameters of the current loop.

[0081] These two steps together constitute the adjustment decision-making and execution phase, in order to adjust the scalar of the audible noise intensity calculated in the previous step. This translates into specific adjustments to the PI controller parameters, thereby forming a closed-loop negative feedback to suppress noise.

[0082] First, the system presets the target energy value under ideal conditions. This value represents the acceptable level of residual audible vibration energy when the system is running quietly; it is typically a small positive number close to zero. It can be set here. A.

[0083] Then, calculate the current weighted auditory vibration energy value. With target value Deviation between This deviation It is the core signal for adjusting the proportional-integral parameters of the drive. This indicates that the current audible noise level exceeds expectations and suppression measures are needed. If This indicates that the system is running well and requires no adjustments; similarly, this applies to other systems.

[0084] Next, based on the deviation Generate the adjustment amount of the proportional-integral parameter; since pure linear adjustment will cause drastic parameter changes when the deviation is large, and simple hard saturation will produce discontinuous control abrupt changes at the saturation boundary, a smooth saturation function can be used:

[0085]

[0086] in, It is the basic adjustment coefficient. It is the maximum allowable adjustment amount, and This is a gain factor (usually set to 1) used to adjust the slope of the transition region of the tanh function. When the value is small, it approaches linear regulation, while... When the value is large, it smoothly approaches the maximum adjustment amount. This avoids controlled mutations and improves system smoothness; and The calibration of parameters such as these typically needs to be performed on an experimental platform: while maintaining stable motor operation, a current disturbance of a specific frequency is gradually injected, and the results are observed. Changes and adjustments Make The response is moderately sensitive; then a large perturbation is applied, and the response is adjusted. This is to ensure that the system does not become unstable due to excessive parameter adjustments.

[0087] Finally, the parameters are updated; the new PI parameters are composed of the old baseline parameters and the calculated adjustment.

[0088]

[0089]

[0090] In this embodiment, Kp is mainly adjusted, therefore It can be set to 0; These are the initial or baseline PI parameters of the system; note that here... The update uses subtraction, meaning that when the noise deviation... When it is positive, the new This will decrease, which aligns with the control principle of suppressing high-frequency oscillations by reducing system bandwidth; updated parameters It will be immediately applied to the proportional-integral controller in the current loop and participate in the current regulation calculation of the next control cycle.

[0091] Thus, the aforementioned process is executed cyclically at a fixed period (e.g., every 1 millisecond), thereby forming a continuously operating adaptive adjustment system.

[0092] For example, the calculated A, Target Value A, then the deviation A.

[0093] Assume the adjustment parameter is: reference proportional gain (V / A), adjustment coefficient (V / A²), maximum adjustment limit (V / A), gain factor .

[0094] Calculate the adjustment amount: ,because ,therefore V / A.

[0095] It is easy to see that the adjustment obtained by using the smooth saturation function (1.766V / A) is less than the maximum value (2.0V / A), but it is still a significant adjustment and the process is smooth.

[0096] Update proportional gain: V / A.

[0097] In the next control cycle, the current loop will operate with a new proportional gain of 8.234, which helps suppress previously detected high-frequency vibrations.

[0098] Example 2

[0099] This embodiment provides a vector control system based on multi-dimensional space. This system is a hardware or hardware-software combined implementation of the method described in Embodiment 1. (Refer to...) Figure 2 This system is typically integrated into a motor controller, servo drive, or frequency converter, and includes:

[0100] Current error acquisition module 101:

[0101] This module is used to perform the aforementioned step S100; the module includes a current sensor (such as a Hall effect sensor) connected to the motor, an analog-to-digital converter (ADC), and a calculation unit that implements coordinate transformation. The calculation unit can be implemented by hardware logic in a digital signal processor (DSP) or a field-programmable gate array (FPGA), and is responsible for performing Clark transformation and Park transformation, and finally outputting the real-time current error signal of the dq axis.

[0102] Vibration Vector Construction Module 102:

[0103] This module is used to execute the aforementioned step S200. It receives the q-axis current error sequence from the current error acquisition module 101. Specifically, this module can be designed in an FPGA as M parallel Gosser algorithm filter cores. Each core has its calculation coefficients hard-coded for the corresponding key frequency points, enabling it to complete one iterative calculation within each sampling clock cycle, thereby achieving extremely high computational efficiency and parallelism. In a DSP-based or high-performance microcontroller (MCU)-based implementation, this module can be a periodically called software function library. After a data window is acquired, the function library calls the Gosser algorithm function M times in a loop to calculate the energy amplitude of M frequency points sequentially. The final output of this module is an M-dimensional array or vector, i.e., a multi-dimensional vibration vector.

[0104] Auditory Energy Assessment Module 103:

[0105] This module is used to execute the aforementioned step S300. Internally, it stores an M-dimensional auditory weighting coefficient vector. This vector can be calculated and stored in memory during system initialization based on preset Gaussian kernel function parameters. This module receives the multi-dimensional vibration vector output by the vibration vector construction module 102, performs the dot product operation on the vectors, and finally outputs a scalar value, namely the weighted auditory vibration energy value. .

[0106] PI parameter adjustment module 104:

[0107] This module is used to execute step S400 in the method flow, and this module receives data from the auditory energy assessment module 103. The value, and compared with the preset target energy value stored in the internal register or memory. Compare and calculate the deviation Subsequently, the module adjusts according to the preset adjustment coefficient. and maximum adjustment limit Perform saturation adjustment logic operations to generate the final PI parameter adjustment amount. .

[0108] PI parameter update module 105:

[0109] This module is used to perform the aforementioned step S500, which reads the reference value of the PI parameter from the internal memory. and the adjustment amount from the PI parameter adjustment module 104 Perform addition and subtraction operations to obtain the updated PI parameters. These new parameter values ​​will be written into the working register of the PI controller module for use in the next control cycle.

[0110] In addition to the core modules mentioned above, the system also includes a standard PI controller module, a space vector pulse width modulation (SVPWM) module, and hardware circuitry for driving the inverter. Together, they form a complete FOC current control closed loop. The system in this application achieves intelligent online optimization of the current loop performance by adding the aforementioned adaptive adjustment path to the standard FOC architecture. Of course, all these modules can be integrated into a single DSP, FPGA, or high-performance MCU chip.

[0111] Example 3

[0112] In this embodiment, in addition to adjusting the proportional-integral parameters after detecting excessive vibration energy in the audible frequency band, an active system identification and feedforward adjustment mechanism is added. This enables the system to not only suppress noise that has already occurred, but also to actively detect changes in the dynamic characteristics of the system, predict potential resonance risks, and adjust the proportional-integral parameters in advance to maintain the robust stability of the system over a wider operating range.

[0113] When the motor is in a non-critical operating phase (e.g., idle, low speed, or low load), the system actively injects a broadband probe signal with extremely low energy but rich spectral density into the control loop. By analyzing the system's response to this probe signal, the system can identify the closed-loop transfer function characteristics under the current proportional-integral parameter settings in real time, especially the resonance peak information related to gain margin and phase margin. Based on this identification result, the system can construct a resonance vulnerability map (i.e., the first map) and, combined with the aforementioned auditory vibration energy, perform multi-objective optimization of the proportional-integral parameters.

[0114] Specifically, the following additional steps may be included:

[0115] S610: Trigger active detection condition judgment.

[0116] In this embodiment, the control system continuously monitors the motor's operating status to determine whether preset triggering conditions are met. These conditions typically include:

[0117] Load conditions: The motor load is below a preset minimum threshold, for example, below 5% of the rated torque.

[0118] Speed ​​conditions: The motor speed is in a low range, such as 10% below the rated speed, or it is stationary.

[0119] Command stability condition: The speed or position command from the upper controller has remained constant or has a very low rate of change over a period of time (e.g., 500 milliseconds).

[0120] The system will enter the active detection phase only when all of the above conditions are met simultaneously. This ensures that the injected weak disturbances will not affect the normal operation of the motor, nor will they be detected by the user through vibration or noise.

[0121] S620: Injects a broadband detection signal into the current control loop.

[0122] In this embodiment, once the triggering condition is met, the system generates a broadband probe signal of a predetermined type and injects it into the current control loop. To obtain high-quality identification results with extremely low injection energy, the preferred probe signal is a pseudo-random binary sequence (PRBS), which has autocorrelation characteristics similar to white noise and a broad, flat spectrum, making it very suitable for system identification.

[0123] For example, the system can generate an m-sequence PRBS signal produced by an 11-stage linear feedback shift register, the amplitude of which is set to be extremely small, for example, equivalent to 0.1% to 0.5% of the q-axis command current. This PRBS signal is digitally and synchronously superimposed on the q-axis voltage command output of the PI controller. Above, that is, the new voltage command is The injection duration is set to be sufficient to cover at least one full cycle of the PRBS signal, for example, 200 milliseconds.

[0124] S630: Synchronous response acquisition and system transmission characteristic analysis.

[0125] In this embodiment, during the entire process of injecting the PRBS probe signal, the system synchronously acquires the error signal of the q-axis current at a high sampling rate. At this time It not only includes the inherent noise of the system, but more importantly, it includes the system's response information to known PRBS inputs.

[0126] After data acquisition, the system uses cross-correlation calculations to extract the system's impulse response; and calculates the injected PRBS signal. With response signal The cross-correlation function between the two can greatly suppress uncorrelated background noise, yielding an approximate impulse response function of the system from q-axis voltage input to q-axis current error output. .

[0127] Subsequently, the system responds to the obtained impulse response function h Perform a Fourier transform to obtain the system's frequency response function. Preferably, the system can directly utilize the vibration vector construction module (i.e., the Gossel algorithm array) already existing in step S200 to process the response signal. In key frequency set The above analysis yields the energy amplitude of the output signal at each frequency point. Since the energy spectrum of the injected PRBS signal at these frequency points is known and approximately constant, the energy spectrum distribution of the response signal directly reflects the gain characteristics of the system transfer function.

[0128] S640: Construct a resonance vulnerability map.

[0129] In this embodiment, the system obtains a set of system gain values ​​at key frequency points. Subsequently, this set of gain values ​​constitutes, as follows: Figure 3 The resonance vulnerability map shown This graph visually represents the system's sensitivity to external disturbances under the current PI parameter settings. The frequency points with higher values ​​in the graph are the system's resonance peaks or quasi-resonance peaks. These peaks mean that if the system happens to encounter a disturbance at that frequency during normal operation, even a small disturbance could be amplified into severe current fluctuations and mechanical vibrations. Therefore, the higher the peak value in the graph, the more vulnerable the system is at that frequency, and the lower its stability margin.

[0130] S650: Performs multi-objective PI parameter optimization with fusion feedback and feedforward.

[0131] In this embodiment, the parameter adjustment no longer depends solely on the current noise level. Instead, it simultaneously considers the resonance vulnerability map, which represents future stability risks. .

[0132] The system constructs a multi-objective cost function. The optimization objective for the PI parameter is to minimize this function:

[0133]

[0134] in:

[0135] It is a weighted auditory vibration energy value based on passive perception, representing the current noise level.

[0136] It is the largest peak in the resonance vulnerability spectrum, representing the maximum potential resonance risk.

[0137] and These are preset weighting coefficients used to balance current performance and future stability. For example, they can be set... .

[0138] The proportional-integral parameter update module is no longer a simple proportional adjustment, but instead employs an optimization strategy based on gradient descent or table lookup. The system can perform small perturbations on the current... Values ​​(e.g., Then, based on the results of active probing, the new cost function is evaluated. The changing trend will determine whether to increase or decrease the next step. until a cost function is found. Minimize the local optimum.

[0139] In this way, the proportional-integral parameters selected by the system can not only keep the current operating state quiet, but also actively avoid resonance regions that are not currently excited but have high potential risks, thereby greatly improving the robustness and stability of the system under unknown operating conditions; the active detection and optimization process can be executed periodically (e.g., once per minute) when the conditions are met, continuously calibrating and optimizing the system model online.

[0140] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0141] It should be noted that the above modules can be implemented by software or hardware. For the latter, they can be implemented in the following ways, but are not limited to: all the above modules are located in the same processor; or, the above modules are located in different processors in any combination.

[0142] Embodiments of the present invention also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to perform the steps in any of the above method embodiments when executed.

[0143] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0144] Embodiments of the present invention also provide an electronic device including a memory and a processor, the memory storing a computer program and the processor being configured to run the computer program to perform the steps in any of the above method embodiments.

[0145] In one exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.

[0146] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0147] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0148] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0149] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0150] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, in essence, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0151] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A vector control method based on multi-dimensional space, characterized in that, include: Obtain the real-time current error signal of the current loop of the motor controller; Based on the real-time current error signal, selective harmonic analysis is performed on preset key frequency points to construct a multidimensional vibration vector characterizing the vibration state of the system, wherein the key frequency points cover the audible frequency band. The multidimensional vibration vector is subjected to a first processing to calculate the perceived vibration energy value; wherein, the first processing of the multidimensional vibration vector includes: weighting and combining each element of the multidimensional vibration vector with the weighting coefficient of the corresponding frequency point to obtain the perceived vibration energy value; Based on the deviation between the perceived vibration energy value and the preset target energy value, a proportional-integral parameter adjustment amount is generated; The proportional-integral controller parameters of the current loop are dynamically updated based on the proportional-integral parameter adjustment.

2. The method according to claim 1, characterized in that, The acquisition of the real-time current error signal of the motor controller current loop includes: Real-time acquisition of three-phase current signals from the motor; A coordinate transformation is performed on the three-phase current signal to obtain the d-axis current component and the q-axis current component in the rotating coordinate system; The d-axis current component and q-axis current component are compared with the d-axis command current and q-axis command current, respectively, to obtain the d-axis current error and q-axis current error, and the q-axis current error is used as the real-time current error signal.

3. The method according to claim 1, characterized in that, After dynamically updating the proportional-integral controller parameters of the current loop, the method further includes: When the preset active detection conditions are met, a broadband detection signal is injected into the current loop; The system response signal generated by the broadband detection signal is collected and analyzed to identify the system gain characteristics of the current loop at key frequency points, so as to obtain the first spectrum. Multi-objective optimization is performed based on the perceived vibration energy value and the first spectrum to update the proportional-integral controller parameters.

4. The method according to claim 3, characterized in that, The multi-objective optimization based on the auditory vibration energy value and the first spectrum includes: The perceived vibration energy value is fused with the maximum gain peak value contained in the first spectrum to obtain a cost function; Adjust the proportional-integral controller parameters to minimize the cost function.

5. A vector control system based on multi-dimensional space, characterized in that, include: The current error acquisition module is used to acquire the real-time current error signal of the current loop of the motor controller; The vibration vector construction module is used to perform selective harmonic analysis on preset key frequency points based on the real-time current error signal in order to construct a multidimensional vibration vector characterizing the vibration state of the system, wherein the key frequency points cover the audible frequency band. The auditory energy assessment module is used to perform a first processing on the multidimensional vibration vector to calculate the auditory vibration energy value; wherein, the first processing on the multidimensional vibration vector includes: weighting and combining each element of the multidimensional vibration vector with the weight coefficient of the corresponding frequency point to obtain the auditory vibration energy value; The parameter adjustment module is used to generate a proportional-integral parameter adjustment amount based on the deviation between the perceived vibration energy value and the preset target energy value. The parameter update module is used to dynamically update the proportional-integral controller parameters of the current loop based on the proportional-integral parameter adjustment amount.

6. The system according to claim 5, characterized in that, The acquisition of the real-time current error signal of the motor controller current loop includes: Real-time acquisition of three-phase current signals from the motor; A coordinate transformation is performed on the three-phase current signal to obtain the d-axis current component and the q-axis current component in the rotating coordinate system; The d-axis current component and q-axis current component are compared with the d-axis command current and q-axis command current, respectively, to obtain the d-axis current error and q-axis current error, and the q-axis current error is used as the real-time current error signal.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program is configured to perform the method described in any one of claims 1 to 4 when executed.

8. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to perform the method as described in any one of claims 1 to 4.

Images