Multi-motor cooperative control method and system based on speed compensation

Abstract

The invention discloses a multi-motor cooperative control method and system based on speed compensation, particularly relates to the technical field of industrial multi-motor control, and is used for solving the problems of compensation misalignment and system oscillation caused by an existing parameter asynchronous drift working condition. Current and voltage signals of each motor are collected in real time, and dynamic resistance and inductance parameters are calculated on line; constructing a parameter variation causal entropy network to quantify causal influence intensity of single machine abnormity on the system, and synchronously generating a system collaborative vulnerability index to realize risk penetration perception; when low-frequency oscillation is detected, harmonic components are extracted, and a sensitivity grading model of parameter mismatch and oscillation frequency is established; a dominant compensation role is distributed based on causal entropy to focus on a key disturbance source, a dynamic amplitude limiting instruction is generated according to collaborative vulnerability to adapt to system stability, and resonance suppression priorities are distributed according to sensitivity levels to optimize resource configuration; and finally, a torque instruction is reconstructed and synchronously issued, so that cross-equipment accurate transmission of compensation energy is realized, and mechanical coupling oscillation is eliminated.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial multi-motor control, and more specifically, to a multi-motor collaborative control method and system based on speed compensation. Background Art

[0002] In an industrial multi-motor collaborative control system, real-time compensation based on speed feedback is the core means to ensure synchronization accuracy. Existing methods rely on preset parameter models such as motor resistance and inductance to generate compensation amounts, and eliminate speed deviations by adjusting torque output. However, in actual operation, motor parameters are continuously drifted due to independent factors such as local temperature rise and magnetic saturation degree differences, and due to the different physical distribution positions and heat dissipation conditions of each motor, the parameter drift shows significant asynchrony.

[0003] The current speed compensation mechanism, due to sticking to the preset parameter model, causes systematic defects under the condition of asynchronous drift of multi-motor parameters: the compensator cannot adapt to the real-time dynamic characteristics of each motor, resulting in response lag or over-compensation; the torque conflict caused by inaccurate compensation is transmitted through mechanical coupling, causing system-level low-frequency oscillation and synchronous error accumulation. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a multi-motor collaborative control method and system based on speed compensation to solve the problems proposed in the above background art.

[0005] To achieve the above object, the present invention provides the following technical solutions: A multi-motor collaborative control method based on speed compensation, comprising: S1. Real-time collect the operating current signals and terminal voltage signals of each motor in the system; S2. According to the operating current signals and terminal voltage signals, online calculate the real-time resistance parameters and real-time inductance parameters of each motor; S3. Based on the real-time resistance parameters and real-time inductance parameters, calculate the parameter mutation causal entropy between each motor through the transfer entropy algorithm, and determine the system collaborative vulnerability based on the Granger causal network analysis; S4. When it is detected that the system has low-frequency oscillation, extract the harmonic components of the operating current signal at the detected low-frequency oscillation frequency, and analyze the sensitivity level of each motor parameter mismatch to the current oscillation frequency; S5. Assign the dominant compensation role according to the parameter mutation causal entropy, generate a dynamic amplitude limiting instruction according to the system collaborative vulnerability, assign the resonance suppression priority according to the sensitivity level, and integrate them into a collaborative compensation strategy; S6. Reconstruct the output torque instruction of the speed compensator according to the collaborative compensation strategy, and synchronously send it to the corresponding motor for execution.

[0006] Further, the operating current signals and terminal voltage signals of each motor in the real-time acquisition system are collected, including: Synchronously sample the instantaneous current values of the three-phase windings of each motor to generate operating current signals; Synchronously sample the instantaneous values of the input voltages at the driving ends of each motor to generate terminal voltage signals.

[0007] Further, the sampling of the operating current signals and terminal voltage signals maintains the same time reference and is updated and stored in continuous cycles.

[0008] Further, based on the operating current signals and terminal voltage signals, the real-time resistance parameters and real-time inductance parameters of each motor are calculated online, including: Establish a parameter calculation model based on the stator voltage equation of the motor. The parameter calculation model represents the terminal voltage signal as a function of the operating current signal and its derivative; Perform central difference calculations on the operating current signals for multiple consecutive sampling periods to obtain current change rate data; Input the terminal voltage signal, operating current signal, and current change rate data into the parameter calculation model to construct an overdetermined system of equations; Iteratively solve the overdetermined system of equations by the recursive least squares method to update the estimated values of the real-time resistance parameters and real-time inductance parameters; When the relative change amount of the estimated value is less than the convergence threshold for a continuously set number of times, output the real-time resistance parameters and real-time inductance parameters.

[0009] Further, based on the real-time resistance parameters and real-time inductance parameters, calculate the parameter mutation causal entropy between each motor by the transfer entropy algorithm, and determine the system collaborative vulnerability based on the Granger causal network analysis, including: Extract the real-time resistance parameters and real-time inductance parameters of each motor according to a set time window to form a parameter time series; Calculate the transfer entropy for the parameter time series of every two motors. The delay order of the transfer entropy is determined according to the minimum point of mutual information; Normalize the transfer entropy value to the parameter mutation causal entropy, and the parameter mutation causal entropy represents the parameter mutation transfer intensity; Construct a directed weighted network based on the parameter mutation causal entropy between all motors. The nodes represent motors, and the edge weights represent the parameter mutation causal entropy; Perform a Granger causality test on the network. When the test statistic exceeds the set significance threshold, add a causal edge; Calculate the variance of the node betweenness centrality of the network as the system collaborative vulnerability.

[0010] Further, when low-frequency oscillations are detected in the system, extract the harmonic components of the operating current signal at the detected low-frequency oscillation frequency, and analyze the sensitivity levels of the mismatches of various motor parameters to the current oscillation frequency, including: Identify the low-frequency components in the operating current signal whose amplitudes exceed a set ratio of the fundamental wave amplitude through spectral analysis. When the frequency of the low-frequency components is within the set low-frequency range, it is determined that low-frequency oscillations exist; When low-frequency oscillations exist, extract the harmonic components with positive and negative set percentage bandwidths centered on the oscillation frequency; Calculate the relative deviation between the measured parameters and the reference parameters of each motor based on the real-time resistance parameters and real-time inductance parameters as the parameter mismatch degree; Establish a linear mapping relationship between the parameter mismatch degree and the harmonic component amplitude, and obtain the sensitivity coefficient through least squares fitting; Divide it into three levels: high-sensitivity level, medium-sensitivity level, and low-sensitivity level according to the absolute value of the sensitivity coefficient.

[0011] Further, allocate the dominant compensation role according to the causal entropy of parameter variation, generate a dynamic amplitude limiting instruction according to the system collaborative vulnerability, allocate the resonance suppression priority according to the sensitivity level, and integrate them into a collaborative compensation strategy, including: Select motors based on the causal entropy of parameter variation sorted from high to low, and allocate the motor with the largest causal entropy of parameter variation as the dominant compensation role; Read the system collaborative vulnerability, generate a dynamic amplitude limiting instruction according to the system collaborative vulnerability, and the amplitude limiting range of the dynamic amplitude limiting instruction is positively correlated with the system collaborative vulnerability; Allocate the resonance suppression priority from the high-sensitivity level to the low-sensitivity level according to the sensitivity level; Fuse the compensation action instruction, dynamic amplitude limiting instruction, and resonance suppression priority instruction of the dominant compensation role to generate a collaborative compensation strategy.

[0012] Further, when the dominant compensation role executes the compensation action, it gives priority to responding to the dynamic amplitude limiting instruction, and the remaining motors execute the compensation action according to the resonance suppression priority.

[0013] Further, reconstruct the output torque instruction of the speed compensator according to the collaborative compensation strategy and synchronously send it to the corresponding motor for execution, including: Analyze the dominant compensation role instruction in the collaborative compensation strategy to determine the compensation torque weight coefficient of the dominant motor; Convert the dynamic amplitude limiting instruction into the upper limit value of torque output, and the upper limit value is used as the amplitude constraint of the torque instruction of all motors; Allocate the torque response timing of each motor according to the resonance suppression priority, and the motor with a higher priority is allocated a front execution time slot; Reconstruct the output torque command of the speed compensator based on the compensation torque weight coefficient, the torque output upper limit value, and the torque response time sequence; Synchronously send the reconstructed output torque command to each motor controller through the real-time communication bus; For the output torque command of the leading motor, the torque weight coefficient is preferentially applied, and the remaining motors are applied in the order of resonance suppression priority.

[0014] On the other hand, the present invention provides a multi-motor cooperative control system based on speed compensation, including: A signal acquisition module for real-time acquisition of the running current signal and terminal voltage signal of each motor in the system; A parameter calculation module for online calculating the real-time resistance parameter and real-time inductance parameter of each motor according to the running current signal and terminal voltage signal; An entropy network analysis module for calculating the parameter mutation causal entropy between each motor through the transfer entropy algorithm based on the real-time resistance parameter and real-time inductance parameter, and determining the system cooperation vulnerability based on the Granger causal network analysis; An oscillation sensitivity analysis module for extracting the harmonic components of the running current signal at the detected low-frequency oscillation frequency and analyzing the sensitivity level of each motor parameter mismatch to the current oscillation frequency when detecting that the system has a low-frequency oscillation; A strategy generation module for allocating the leading compensation role according to the parameter mutation causal entropy, generating a dynamic amplitude limiting command according to the system cooperation vulnerability, allocating the resonance suppression priority according to the sensitivity level, and integrating them into a cooperative compensation strategy; An instruction reconstruction module for reconstructing the output torque command of the speed compensator according to the cooperative compensation strategy and synchronously sending it to the corresponding motor for execution.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By real-time calculating the dynamic resistance and inductance parameters of each motor, breaking through the rigid constraints of the traditional preset parameter model, eliminating the problem of compensation reference inaccuracy caused by parameter asynchronous drift, constructing a parameter mutation causal entropy network based on the transfer entropy algorithm, accurately quantifying the influence intensity of single-motor parameter drift on the overall system, synchronously generating a system cooperation vulnerability index, realizing a penetrative perception from local anomalies to system risks, when detecting a low-frequency oscillation, through the sensitivity mapping of harmonic components and parameter mismatch, establishing an equipment-level oscillation response classification model, and forming a direct association between fault tracing and suppression targets.

[0016] 2. The collaborative compensation strategy integrates a triple complementary decision-making mechanism: The responsibility allocation dominated by the causal entropy of parameter mutation focuses on key disturbance sources. The dynamic amplitude limiting driven by the system collaborative vulnerability is adapted to the system stability margin in real time. The priority of resonance suppression in the sensitivity level ranking realizes the spatio-temporal optimal allocation of compensation resources. The torque command reconstruction converts the electrical compensation strategy into mechanical regulation actions. Through the binding of weight coefficients and the decoupling of execution timings, it ensures the accurate transmission of compensation energy between cross-devices. The synchronous distribution mechanism eliminates the distributed execution timing deviation. The role rotation strategy avoids the single-point overload risk. Finally, it realizes the global dynamic optimization of speed compensation in multi-motor collaborative control. Description of the Drawings

[0017] Figure 1 is a flowchart of the multi-motor collaborative control method based on speed compensation of the present invention; Figure 2 is a schematic structural diagram of the multi-motor collaborative control system based on speed compensation of the present invention. Detailed Embodiments

[0018] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0019] Embodiment 1: Figure 1 The multi-motor collaborative control method based on speed compensation of the present invention is given, including: S1. Real-time collect the running current signals and terminal voltage signals of each motor in the system; S2. According to the running current signals and terminal voltage signals, online calculate the real-time resistance parameters and real-time inductance parameters of each motor; S3. Based on the real-time resistance parameters and real-time inductance parameters, calculate the causal entropy of parameter mutation between each motor through the transfer entropy algorithm, and determine the system collaborative vulnerability based on the Granger causal network analysis; S4. When it is detected that there is a low-frequency oscillation in the system, extract the harmonic components of the running current signal at the detected low-frequency oscillation frequency, and analyze the sensitivity level of each motor parameter mismatch to the current oscillation frequency; S5. Allocate the dominant compensation role according to the causal entropy of parameter mutation, generate a dynamic amplitude limiting command according to the system collaborative vulnerability, allocate the priority of resonance suppression according to the sensitivity level, and integrate them into a collaborative compensation strategy; S6. Reconstruct the output torque command of the speed compensator according to the collaborative compensation strategy, and synchronously send it to the corresponding motor for execution.

[0020] S1. The running current signals and terminal voltage signals of each motor in the real-time acquisition system are collected as follows: The running current signals and terminal voltage signals of each motor in the real-time acquisition system are achieved in the following way: Use current sensors to measure the currents of the three-phase windings of each motor respectively. The current sensors adopt the closed-loop Hall effect principle, and their measurement bandwidth is determined according to the highest operating frequency of the motor. The lower limit value of the measurement bandwidth is twice the highest electrical frequency of the motor. For example, when the highest electrical frequency of the motor is 500 Hz, the measurement bandwidth is set to be greater than 1000 Hz. The analog current signals output by the current sensors are subjected to anti-aliasing filtering through a signal conditioning circuit. The cut-off frequency of the anti-aliasing filtering is set to be half of the motor control sampling frequency according to the Nyquist sampling theorem. The filtered signals are digitized to sample the instantaneous current values at fixed time intervals through an analog-to-digital converter. The sampling of the instantaneous current values of the three-phase windings of each motor is triggered synchronously by the same controller, ensuring that the time deviation of the three-phase current sampling moments is less than 1% of the controller instruction cycle, so as to generate time-aligned running current signals.

[0021] The acquisition of the terminal voltage signals is achieved by directly connecting a differential voltage detection circuit to the power output terminal of the motor driver to measure the potential difference between the three-phase input terminals of the motor relative to the negative terminal of the DC bus. After the voltage signals are subjected to voltage division attenuation and isolation amplification, they are input to the analog-to-digital converter for synchronous sampling. The trigger signal for voltage sampling and the current sampling use the synchronous pulse issued by the same controller, so that the running current signals and terminal voltage signals of the same motor have exactly the same time stamp in each sampling period. To eliminate the interference of switching noise, a second-order low-pass filter is inserted before the voltage signals enter the analog-to-digital converter. The cut-off frequency of this filter needs to meet both requirements of being higher than ten times the fundamental frequency of the motor and lower than half of the switching frequency. For example, when the fundamental frequency of the motor is 50 Hz and the switching frequency is 8 kHz, the cut-off frequency is set to 3 kHz.

[0022] The sampling of the running current signals and terminal voltage signals maintains the same time reference through the following mechanism: A high-precision timer is set inside the controller. The timer clock source adopts a temperature-compensated crystal oscillator, and the frequency stability of the crystal oscillator is better than five parts per million. At the beginning of each sampling period, the timer generates a global synchronous interrupt signal, which is transmitted to the current sampling modules and voltage sampling modules of all motors simultaneously through a hardware trigger line to trigger the synchronous start of conversion of the analog-to-digital converters of each channel. When the data after analog-to-digital conversion is written into the dual-port memory, the memory address pointer is uniformly updated by the timer interrupt service program to ensure that the data of the running current signals and terminal voltage signals of all motors have the same time stamp serial number.

[0023] The sampled data is updated and stored in continuous cycles, which is implemented by a circular buffer mechanism: a data storage area with a fixed length is allocated in the controller memory, and the storage area is managed according to the first-in, first-out principle. The depth of the storage area is set according to the dynamic response requirements of the system, specifically based on the number of sampling points required to cover ten times the electromagnetic time constant of the motor. For example, when the electromagnetic time constant of the motor is 10 milliseconds and the sampling period is 100 microseconds, the depth of the storage area is set to 1000 sets of data. When the data is updated, the newly sampled running current signal and terminal voltage signal are written into the current pointer position of the buffer in chronological order, and the pointer automatically increments to the next storage unit. When the pointer reaches the end of the storage area, it automatically returns to the starting address, forming a circular overwrite.

[0024] The selection of the current sensor needs to meet the following technical specifications: the lower limit of the measurement range is 150% of the rated current of the motor, and the upper limit is 200%; the linearity error is less than 0.5% of the full scale; the response time is less than one-fifth of the sampling period. For example, for a motor with a rated current of 50 amperes, the measurement range of the current sensor is set to 0 to 100 amperes. In the signal conditioning circuit, the anti-aliasing filter is designed using a Butterworth low-pass filter topology, with a passband ripple of less than 0.1 dB and a stopband attenuation of more than 60 dB, and the group delay characteristic is optimized through a phase compensation circuit to ensure that the phase difference measurement error of each phase current signal is less than 0.1 degree.

[0025] The parameter configuration of the analog-to-digital converter follows the following rules: the sampling resolution is not less than 12 bits to meet the control accuracy requirements; the sampling rate is at least twenty times the fundamental frequency of the motor. For example, when the fundamental frequency of the motor is 100 Hz, the sampling rate is set to not less than 2 kHz. The sampling synchronous trigger uses a dedicated hardware trigger line, and the signal transmission delay is controlled within 1 nanosecond through equal-length wiring. When storing data, the running current signal of each sampling point contains the instantaneous values of the three-phase currents, and the terminal voltage signal contains the instantaneous values of the three-phase voltages. The storage format uses 32-bit floating-point numbers, and the time stamp information uses a 64-bit microsecond-level time stamp.

[0026] A data protection mechanism is set for the continuous cycle storage area of the running current signal and the terminal voltage signal: when the background program reads historical data, the storage unit corresponding to this cycle is locked to prohibit writing. The storage area management adopts a double-pointer structure at the head and tail. The head pointer indicates the position of the latest data, and the tail pointer indicates the position of the oldest data. If a storage area overflow is detected, a data compression algorithm is started: the sampling rate of the earliest 50% of the historical data is reduced to half of the original sampling rate using linear interpolation, and a data compression flag bit is set for subsequent processing identification.

[0027] S2. According to the running current signal and the terminal voltage signal, the real-time resistance parameters and real-time inductance parameters of each motor are calculated online, and the specific implementation is as follows: The specific implementation process of online calculating the real-time resistance parameters and real-time inductance parameters of each motor according to the operating current signal and terminal voltage signal is as follows: The stator voltage equation of the permanent magnet synchronous motor in the rotating coordinate system is used as the basis of the parameter calculation model. This model decomposes the terminal voltage signal into the d-axis component and the q-axis component. The d-axis component is equal to the d-axis component of the operating current signal multiplied by the real-time resistance parameter plus the derivative of the d-axis component of the operating current signal multiplied by the real-time inductance parameter minus the electrical angular velocity multiplied by the real-time inductance parameter multiplied by the q-axis component of the operating current signal. The q-axis component is equal to the q-axis component of the operating current signal multiplied by the real-time resistance parameter plus the derivative of the q-axis component of the operating current signal multiplied by the real-time inductance parameter plus the electrical angular velocity multiplied by the real-time inductance parameter multiplied by the d-axis component of the operating current signal plus the permanent magnet flux linkage multiplied by the electrical angular velocity. The electrical angular velocity is calculated by multiplying the mechanical angular velocity measured by the motor rotor position sensor by the number of pole pairs of the motor, and the mechanical angular velocity is obtained in real time by the incremental encoder installed at the motor shaft end.

[0028] The operation process of obtaining the current change rate data by performing central difference calculation on the operating current signals of consecutive multiple sampling periods is as follows: Read the operating current signals at the current moment and the adjacent two sampling periods from the data storage area, and calculate independently for each phase of the three-phase current. For the calculation of the current change rate at the intermediate sampling point k, take the difference between the instantaneous value of the current at the previous sampling point k - 1 and the instantaneous value of the current at the next sampling point k + 1, and divide the difference by twice the sampling period time to obtain the current change rate. For example, when the sampling period is 100 microseconds, the current change rate is equal to the instantaneous value of the operating current signal at the sampling point k + 1 minus the instantaneous value of the operating current signal at the sampling point k - 1 and then divided by 200 microseconds. The calculation process is executed by a 32-bit floating-point arithmetic unit, and the data buffer is updated after each calculation.

[0029] The steps of inputting the terminal voltage signal, operating current signal, and current change rate data into the parameter calculation model to construct an overdetermined system of equations include: Select the data of 20 consecutive sampling points to form an overdetermined system of equations, and each sampling point corresponds to two equations respectively describing the d-axis and q-axis relationships. Each equation contains five known quantities: the instantaneous value of the d-axis component of the terminal voltage signal at the current sampling point, the instantaneous value of the q-axis component of the terminal voltage signal, the instantaneous value of the d-axis component of the operating current signal, the instantaneous value of the q-axis component of the operating current signal, the instantaneous value of the electrical angular velocity, and two unknown parameters: the real-time resistance parameter and the real-time inductance parameter. When constructing the system of equations, the data points during the period of severe current change are automatically excluded, and the determination condition for severe change is that the absolute value of the current change rate exceeds three times the rated current change rate of the motor.

[0030] The implementation method of updating the parameter estimation value by iteratively solving the overdetermined equation set using the recursive least squares method is as follows: Initialize the initial estimation values of the real-time resistance parameter and the real-time inductance parameter as the nominal values at the time of motor factory shipment, and initialize the covariance matrix as 1000 times the identity matrix. Each time new sampled point data arrives, calculate the prediction residual of the new data vector and the current parameter estimation value, calculate the gain matrix based on the current covariance matrix and the data vector, update the parameter estimation value using the gain matrix and the residual, and finally update the covariance matrix. During the iteration process, the forgetting factor is set as a dynamically adjustable parameter, with an initial value set to 0.98, and it is automatically adjusted according to the residual change rate.

[0031] The determination logic when the relative change amount of the parameter estimation value is less than the convergence threshold for a continuous set number of times is as follows: Calculate the relative change amount between the current estimation value and the previous estimation value after each parameter update. The relative change amount is defined as the absolute value of the difference between the two estimation values divided by the absolute value of the previous estimation value. Set the convergence threshold to 0.001 and the continuous satisfaction number to 5 times. For example, when the relative change amount of the real-time resistance parameter in 5 consecutive iterations is less than 0.001, it is determined to converge. After convergence, lock the parameter estimation value and output it to the data bus, and at the same time reset the iteration counter. If convergence is not achieved after 200 iterations, take the median of the results of the last 50 iterations as the output value.

[0032] Add a data preprocessing link in the calculation process of the current change rate: Perform three-point moving average filtering on the original operating current signal, and the filtering window covers the current sampling point and one sampling point before and after it. The filtered current value is used for central difference calculation to effectively suppress high-frequency measurement noise. When an abnormal jump in the current change rate data is detected, the jump determination threshold is five times the difference in the current change amount between adjacent sampling points, and the standby sampling point is automatically enabled to recalculate the current change rate.

[0033] The numerical stability measure for updating the covariance matrix in the recursive least squares method is as follows: Monitor the condition number of the covariance matrix in real time. When the condition number is greater than 1000000, add a perturbation value to the main diagonal elements of the covariance matrix, and the size of the perturbation value is 1% of the average value of the diagonal elements. The adaptive adjustment rule for the forgetting factor is as follows: Calculate the change rate of the root mean square value of the residuals in the last 10 iterations. When the change rate is continuously positive, reduce the forgetting factor by 0.001 until the lower limit value of 0.95; when the change rate is continuously negative, increase the forgetting factor by 0.001 until the upper limit value of 0.99.

[0034] Perform validity verification before parameter output: The real-time resistance parameter value must be within the resistance change range corresponding to the allowable operating temperature of the motor, and the allowable range is 80% to 120% of the nominal value; the real-time inductance parameter value must be within the allowable interval of the motor magnetic saturation characteristic, and the allowable range is 60% to 150% of the nominal value. If the parameter exceeds the allowable range, the average value of the last 10 valid iteration results is used for substitution, and the parameter exception status flag is triggered.

[0035] The overdetermined equation system is constructed using matrix block technology: The data of 20 consecutive sampling points are arranged in a 40-row and 2-column observation matrix in chronological order. The first column of the observation matrix is the d-axis component and q-axis component of the operating current signal of all sampling points, and the second column is the corresponding current change rate data. The terminal voltage signal vector consists of 40 elements, including the instantaneous values of the d-axis component and q-axis component of the terminal voltage signal arranged alternately. Invalid data points are automatically skipped during equation construction, and the criterion for determining invalid points is that the absolute value of the current change rate exceeds twice the maximum allowable current change rate of the motor.

[0036] The permanent magnet flux linkage parameter in the parameter solution model uses the offline calibration value, and the calibration method is to measure the back electromotive force constant under the no-load state of the motor and divide it by the electrical angular velocity. When converting the electrical angle, the mechanical angular velocity is multiplied by the number of pole pairs of the motor to obtain the electrical angular velocity, and the number of pole pairs of the motor is fixed as a set value according to the motor model. For example, the number of pole pairs corresponding to a 4-pole motor is 2. The permanent magnet flux linkage value required for calculating the back electromotive force term is stored in the non-volatile memory and input as a known constant during the parameter solution process.

[0037] S3. Based on the real-time resistance parameter and real-time inductance parameter, calculate the parameter mutation causal entropy between each motor through the transfer entropy algorithm, and determine the system collaborative vulnerability based on the Granger causal network analysis. The specific implementation is as follows: The specific implementation process of extracting the real-time resistance parameter and real-time inductance parameter of each motor to form a parameter time series according to the set time window is as follows: Set a time window with a fixed duration. The length of the time window is determined according to the thermal time constant of the motor, and the value is one-fifth to one-tenth of the thermal time constant. For example, when the thermal time constant of the motor is 30 minutes, the length of the time window is set to 3 minutes. Extract the real-time resistance parameter and real-time inductance parameter of all sampling points within this time window in the order of the sampling period from the data storage area. Each sampling point corresponds to a set of parameter values. The real-time resistance parameter and real-time inductance parameter respectively form independent time series, and the length of the time series is equal to the number of sampling points included in the time window. When the sampling period is 100 milliseconds, the 3-minute time window corresponds to 1800 sampling point data. The time series data storage is managed using a circular buffer, and new data overwrites the earliest historical data.

[0038] The operation process of calculating transfer entropy for the parameter time series of every two motors is as follows: Select the same type of parameter time series of the target motor and the source motor as the input data. First, calculate the mutual information function between the source parameter time series and its own delay series. The delay order increases from the first order to the maximum set order of 50 orders. The delay order corresponding to the first minimum point of the mutual information function is determined as the delay order for transfer entropy calculation. The transfer entropy calculation is implemented using the histogram probability estimation method: Dynamically divide the parameter value range into a set number of equal-width intervals, and the number of intervals is the integer value obtained by taking the square root of the time series length. For example, when the time series contains 1000 sampling points, 31 intervals are divided. Statistically calculate the joint distribution probability of the historical state of the source parameter, the current state and the historical state of the target parameter. The transfer entropy value is equal to the joint entropy of the historical state of the source parameter and the current state of the target parameter minus the conditional entropy of the historical state of the source parameter minus the conditional entropy of the historical state of the target parameter itself.

[0039] The process of normalizing the transfer entropy value into the parameter mutation causal entropy is as follows: Calculate the Shannon entropy of the source parameter time series. The Shannon entropy is equal to the negative sum of the probability of each parameter value multiplied by the logarithm of that probability to the base 2. The parameter mutation causal entropy is equal to the transfer entropy value divided by the Shannon entropy of the source parameter. The value range of the normalized parameter mutation causal entropy is between 0 and 1. The larger the value, the stronger the parameter mutation transfer intensity. For example, when the transfer entropy value is 0.8 bits and the Shannon entropy of the source parameter is 2.0 bits, the calculation result of the parameter mutation causal entropy is 0.4. Perform a validity check on the normalization result. When the Shannon entropy is less than 0.01 bits, it is determined as invalid data and automatically replaced with the previous valid window value.

[0040] The steps of constructing a directed weighted network based on the parameter mutation causal entropy between all motors include: Map each motor in the system to a network node. For any two motor nodes, if the parameter mutation causal entropy from the source motor to the target motor is greater than the set effective threshold, add a directed edge pointing from the source node to the target node. The weight value of the directed edge is set to the parameter mutation causal entropy value in the corresponding direction. The effective threshold is dynamically adjusted according to the network density requirement. The initial value is set to 0.15 and automatically adjusted according to the network connectivity: When the number of network edges is less than 1.5 times the number of nodes, the threshold is reduced by 0.02; when the number of edges is more than 3 times the number of nodes, the threshold is increased by 0.02. For example, in a 10-motor system, the initial effective threshold is 0.15. When it is detected that the number of edges is less than 15, the threshold is reduced to 0.13.

[0041] The execution method for performing Granger causality test on the network is as follows: Perform Granger causality test on the time series of motor parameters corresponding to each established directed edge. An autoregressive model of the target parameter and an extended regression model including the source parameter are established. The model order is determined by the Bayesian information criterion, and the order search range is from 1 to 10. The least squares method is used to fit the model coefficients, and an L2 regularization term is added during the least squares solution. The regularization coefficient is set to 1% of the residual variance. Calculate the sum of squared residuals of the two models, construct the F statistic, and the F statistic is equal to the difference between the sum of squared residuals of the two models divided by the sum of squared residuals of the extended model and then multiplied by the degree of freedom ratio. When the value of the F statistic exceeds the set significance threshold, confirm that this directed edge is a valid causal edge. The significance threshold takes the critical value of the F distribution at the 95% confidence level. For example, when the degrees of freedom are (5, 1000), it takes 2.21.

[0042] The process of calculating the variance of the betweenness centrality of network nodes as the system collaborative vulnerability is as follows: First, calculate the betweenness centrality of each node. The betweenness centrality is equal to the proportion of the shortest paths between all node pairs in the network passing through this node. The Floyd-Warshall algorithm is used to calculate the shortest paths: Initialize the path matrix. The path length between directly connected nodes is taken as the reciprocal of the edge weight value, and the path length between non-directly connected nodes is set to a large value, and the large value is taken as 100 times the sum of the reciprocals of all edge weights. Iteratively update the path matrix until convergence. Count the number of shortest paths passing through each node, and the betweenness centrality is equal to this number divided by the total number of all node pairs. After the betweenness centralities of all nodes form a set, calculate the variance value of this set as the system collaborative vulnerability. The larger the variance value, the worse the system collaborative stability. For example, when the variance value is 0.25, it indicates that the system has a single point of failure risk.

[0043] The preprocessing of the parameter time series includes: performing first-order difference operations on the real-time resistance parameter and the real-time inductance parameter respectively to eliminate the non-stationary trend. The difference operation formula is the current value minus the value of the previous sampling point. Perform Z-score standardization on the differenced parameter values: subtract the time series mean and then divide by the standard deviation. The standardized parameter values form a new time series for subsequent calculations. When the length of the time series is insufficient, the time window is automatically extended, and the extension step is half of the original window length. The maximum number of extensions is set to 3 times.

[0044] The probability estimation in transfer entropy calculation adopts the adaptive interval partitioning technique: Dynamically adjust the interval width according to the distribution characteristics of the parameter values so that each interval contains approximately equal numbers of samples. The specific implementation is as follows: After arranging the parameter values in ascending order, divide them equally into the target number of intervals, and each quantile is used as the interval boundary. The interval boundary smoothing process uses a three-point moving average filter to eliminate boundary mutations. When performing probability density estimation, Gaussian kernel smoothing is performed on the histogram, and the kernel width is set to one-tenth of the interval width.

[0045] Model residual diagnosis in Granger causality test: Calculate the residual autocorrelation function of the extended regression model. When the absolute value of the first-order autocorrelation coefficient is greater than 0.2, it is determined that there is autocorrelation and the model order is automatically increased. The maximum allowable order is extended to twice the initial order. If the requirements are still not met, a reliability warning flag is marked for this edge.

[0046] Edge validity verification mechanism during network construction: When the parameter mutation causal entropy is greater than the effective threshold but less than the confidence threshold, three time windows need to be continuously verified. If two of the three windows meet the conditions, the directed edge is confirmed to be valid. The confidence threshold is set to 1.5 times the effective threshold. For example, when the effective threshold is 0.15, the confidence threshold is 0.225. When a conflict result is detected during the verification process, start the statistical verification based on Bootstrap resampling: Generate 100 groups of resampling sequences to recalculate the parameter mutation causal entropy. When more than 90% of the results are greater than the effective threshold, it is confirmed to be valid.

[0047] Path weight optimization in betweenness centrality calculation: Convert the edge weight into an information transfer efficiency value. The conversion formula is 1.5 minus the parameter mutation causal entropy. The converted weight value ranges from 1.0 to 1.5. The greater the weight, the higher the information transfer efficiency. The shortest path calculation adopts the efficiency-first principle, and the path length is equal to the sum of the converted weights of the passing edges.

[0048] Output calibration processing of system collaborative vulnerability: When the number of network nodes is less than 4, the variance of node degree centrality is automatically used to replace the variance of betweenness centrality. The degree centrality is defined as the weighted sum of the in-degree and out-degree of the node. The in-degree weight is set to 0.7, and the out-degree weight is set to 0.3. The variance calculation result is multiplied by the scale compensation coefficient, and the scale compensation coefficient is equal to the square root of the number of nodes divided by the reference value 4.

[0049] In this step S3, through the dual analysis mechanism that combines transfer entropy and Granger causal network, it can capture the hidden fault transmission path of the multi-motor system more accurately compared to a single algorithm. Specifically: By using the normalization processing of transfer entropy, the influence of parameter dimension differences and entropy value scales is eliminated, making the transmission intensity of parameter mutations comparable between different motors, and avoiding the misjudgment of non-linear relationships by traditional correlation coefficients. First, an initial network is constructed based on information theory, and then spurious correlation edges (such as false associations caused by common environmental interference) are removed through Granger causality tests, improving the reliability of causal edges. Compared with a single Granger test, it reduces the dependence on data stationarity and linear assumptions. The variance of betweenness centrality (instead of the conventional degree centrality) is selected to characterize the system vulnerability, which can sensitively identify the cascading risks caused by the failure of key hub nodes and is more in line with the topological vulnerability characteristics of the multi-motor collaborative scenario. While ensuring the universality of the algorithm (such as the transfer entropy dealing with non-linearity), by combining statistical verification with physical characteristics (such as setting a window for the thermal time constant), the computational efficiency and diagnostic accuracy are effectively balanced.

[0050] S4. When a low-frequency oscillation is detected in the system, extract the harmonic components of the operating current signal at the detected low-frequency oscillation frequency, and analyze the sensitivity level of each motor parameter mismatch to the current oscillation frequency. The specific implementation is as follows: The specific implementation process of identifying the low-frequency components in the operating current signal whose amplitude exceeds a set proportion of the fundamental wave amplitude through spectrum analysis and determining the existence of low-frequency oscillation when the frequency of the low-frequency component is within the set low-frequency range is as follows: Perform windowed Fourier transform processing on the operating current signal. The window function type is selected as the Hanning window to reduce spectrum leakage, and the window length is set to an integer multiple of the fundamental wave period to ensure complete cycle sampling. Calculate the amplitude of each frequency component in the spectrum, where the fundamental wave amplitude specifically refers to the amplitude of the power frequency component. The set proportion is determined according to the power system stability requirements, and the specific value range is 20% to 40% of the fundamental wave amplitude, which is obtained through statistical analysis of historical fault data. The set low-frequency range is determined according to the motor type. For synchronous motors, it is usually taken as 0.5 Hz to 5 Hz, and this range covers the typical low-frequency oscillation frequency band of the motor. When it is detected that the amplitude of a certain low-frequency component exceeds the set proportion and the frequency is within the set low-frequency range, it is determined that the system is in a low-frequency oscillation state. For example, when the fundamental wave amplitude is 100 amperes and the set proportion is 30%, the amplitude of the low-frequency component exceeding 30 amperes triggers the determination.

[0051] When there is low-frequency oscillation, the operation process of extracting harmonic components with positive and negative set percentage bandwidths centered around the oscillation frequency is as follows: The set percentage bandwidth is dynamically adjusted according to the oscillation mode characteristics, usually taking 10% to 20% of the oscillation frequency. A smaller bandwidth is taken for high-frequency oscillation, and a larger bandwidth is taken for low-frequency oscillation. A band-pass digital filter is constructed to achieve harmonic extraction. The center frequency of the filter is set to the oscillation frequency, and the bandwidth is equal to twice the product of the oscillation frequency and the set percentage bandwidth value. The running current signal is subjected to real-time filtering processing to extract the harmonic component sequence in the target frequency band. The filtered signal calculates the instantaneous amplitude through Hilbert transform to form the harmonic component amplitude time series, and the sampling interval is consistent with the original current signal. For example, when the oscillation frequency is 2 Hz and the set percentage bandwidth is 15%, the harmonic components in the frequency band range of 1.7 Hz to 2.3 Hz are extracted.

[0052] The process of calculating the relative deviation between the measured parameters and the reference parameters of each motor based on the real-time resistance parameters and real-time inductance parameters as the parameter mismatch degree is as follows: The reference parameters adopt the rated parameters of the motor or the average value of the parameters in the last 10 sampling periods under the system stable state. The real-time resistance parameters and real-time inductance parameters are obtained from the parameter storage area, and the sampling timestamps are strictly aligned with the harmonic component amplitude time series. The parameter mismatch degree calculation formula is (measured parameter - reference parameter) / reference parameter * 100%, and the calculation result is a dimensionless percentage value. The absolute values of the resistance parameter mismatch degree and the inductance parameter mismatch degree are calculated respectively, and the two are weighted and summed to obtain the comprehensive parameter mismatch degree, where the resistance weight value range is 0.5 to 0.7, and the inductance weight value range is 0.3 to 0.5. The specific weights are determined through sensitivity analysis according to the motor type. For example, when the measured resistance value is 10.0 ohms and the reference value is 9.5 ohms, the resistance mismatch degree is calculated as 5.26%.

[0053] The steps of establishing a linear mapping relationship between the parameter mismatch degree and the harmonic component amplitude and obtaining the sensitivity coefficient through least squares fitting include: Collecting the data pairs of the parameter mismatch degree and the harmonic component amplitude at all sampling points within the current time window to form a sample set. Constructing a linear regression model, with the independent variable being the parameter mismatch degree and the dependent variable being the harmonic component amplitude. Solving the regression coefficients through the least squares method, and this regression coefficient is the sensitivity coefficient. An outlier rejection mechanism is adopted during the fitting process. When the residual exceeds 3 times the sample standard deviation, it is determined as an outlier and discarded. The fitting process is repeated until the residual change rate of two consecutive iterations is less than 5%. The unit of the sensitivity coefficient is unified as amperes per percentage, indicating the change amount of the harmonic amplitude caused by a 1% change in the parameter mismatch degree. For example, the fitted sensitivity coefficient is 1.2 amperes per percentage, indicating that for every 1% increase in the parameter mismatch degree, the harmonic amplitude increases by an average of 1.2 amperes.

[0054] The execution method for dividing into three levels of high - sensitivity level, medium - sensitivity level, and low - sensitivity level according to the absolute value of the sensitivity coefficient is as follows: Set two classification thresholds. The value range of the first threshold is from 0.4 to 0.6 amperes per percentage, and the value range of the second threshold is from 0.15 to 0.25 amperes per percentage. The specific thresholds are dynamically adjusted according to the system operation state. When the system is in a critically stable state, lower thresholds are taken to improve the detection sensitivity. When the absolute value of the sensitivity coefficient is greater than the first threshold, it is classified as the high - sensitivity level; when the absolute value of the sensitivity coefficient is less than the first threshold but greater than the second threshold, it is classified as the medium - sensitivity level; when the absolute value of the sensitivity coefficient is less than the second threshold, it is classified as the low - sensitivity level. The sensitivity coefficients corresponding to the comprehensive parameter mismatch degrees of each motor are classified respectively to generate a sensitivity level list. For example, if the sensitivity coefficient of a certain motor is 0.6 amperes per percentage, when the first threshold is 0.5, it is classified as the high - sensitivity level.

[0055] Parameter alignment and time synchronization processing: The time series of the harmonic component amplitude and the time series of the parameter mismatch degree are precisely matched in time stamps, and the maximum allowable time deviation is 1 / 2 of the sampling period. When the time deviation exceeds the limit, cubic spline interpolation is used for data resampling to align the time points. The interpolated data points are given lower weights in the least - squares fitting, and the weight coefficient is taken as 0.5 times the weight of the measured data. This weight coefficient is determined through error propagation analysis.

[0056] Sensitivity coefficient verification mechanism: Conduct a statistical significance test on the fitted sensitivity coefficient, calculate the t - statistic and compare it with the critical value at the set significance level. The t - statistic is equal to the sensitivity coefficient divided by the standard error, and the standard error is calculated through the sum of squared residuals and the number of samples. When the absolute value of the t - statistic is less than the critical value, it is determined that the sensitivity coefficient is not significant, and the corresponding sensitivity level is marked as the to - be - confirmed state. The significance level is usually taken as 5%, and the critical value is determined by looking up the t - distribution table according to the degrees of freedom. The degrees of freedom is equal to the number of samples minus 2. For example, when the number of samples is 50, the degrees of freedom is 48, and the corresponding critical value is 2.01.

[0057] Sensitivity level output format: The sensitivity level of each motor is output bound to the sensitivity coefficient, including three - level classifications corresponding to the sensitivity coefficients of the resistance parameter mismatch degree, the inductance parameter mismatch degree, and the comprehensive sensitivity coefficient. The output results are sorted in descending order of the comprehensive sensitivity level, and the high - sensitivity - level motors are prioritized. For the sensitivity coefficients in the to - be - confirmed state, an additional reliability flag is attached to the output, and this flag is a boolean variable.

[0058] Abnormal condition handling: When multiple oscillation frequencies are detected, the dominant frequency is selected by sorting according to the oscillation energy magnitude. The oscillation energy is calculated as the square of the amplitude multiplied by the bandwidth. Sensitivity analysis is performed separately on the three oscillation frequencies with the largest energy, and the analysis results corresponding to each oscillation frequency are output independently and marked with frequency identifiers. When the energy difference between two oscillation frequencies is less than 10%, the double dominant frequency identifier is marked simultaneously.

[0059] Parameter mismatch degree recalculation mechanism: When a sudden change occurs in the harmonic component amplitude, the recalculation of the parameter mismatch degree is triggered. The sudden change determination condition is that the difference between the current amplitude and the average value of the previous 10 sampling points exceeds 3 times the standard deviation. When recalculating, a sliding window is used to update the reference parameters, and the window length is shortened to 1 / 2 of the original window.

[0060] S5. Assign the dominant compensation role according to the causal entropy of parameter variation, generate a dynamic amplitude limiting instruction according to the system collaborative vulnerability, assign the resonance suppression priority according to the sensitivity level, and integrate them into a collaborative compensation strategy. The specific implementation is as follows: Select the motor based on the causal entropy of parameter variation sorted from high to low. The specific implementation process of assigning the motor with the largest causal entropy of parameter variation as the dominant compensation role is as follows: directly read the calculated causal entropy values of each motor from the system parameter storage area. This value has been obtained through the time-varying correlation analysis of the parameter mismatch degree and the harmonic component amplitude in the previous steps. Perform a descending order operation on the causal entropy values of all motors, and select the single motor with the largest value in the sorting result as the dominant compensation role. When the difference in the causal entropy values of two or more motors is less than the set tolerance of 5%, the motor with a larger comprehensive parameter mismatch degree is selected as the dominant compensation role. The dominant compensation role is given the core compensation function, and the lower limit of the total system compensation amount it undertakes is 60%. For example, when the causal entropy of motor A's parameter variation is 0.85, that of motor B is 0.83, and the set tolerance is 5%, since the difference of 2.4% is less than 5%, the motor with a larger comprehensive parameter mismatch degree is selected at this time.

[0061] Read the system collaborative vulnerability, and the specific operation process of generating a dynamic amplitude limiting instruction according to the system collaborative vulnerability is as follows: obtain the quantified system collaborative vulnerability value in real time from the system status database. This value has been calculated through the multi-motor impedance coupling characteristic analysis in the previous steps. Construct a dynamic amplitude limiting mapping rule table, map the system collaborative vulnerability value range to the amplitude limiting ratio, and the mapping relationship is a monotonically increasing function. The amplitude limiting ratio is defined as the percentage of the maximum allowable output current value to the reference current value, and its value increases linearly with the increase of the system collaborative vulnerability. The dynamic amplitude limiting instruction output is the maximum allowable compensation current value, and the calculation method is the reference compensation current multiplied by the amplitude limiting ratio. For example, when the system collaborative vulnerability is 0.7, the corresponding amplitude limiting ratio is 80%. When the reference compensation current is 100 amperes, an 80-ampere amplitude limiting instruction is generated.

[0062] The specific processing procedure for allocating resonance suppression priorities from the high-sensitivity level to the low-sensitivity level is as follows: Obtain the classified results of the sensitivity levels of each motor that have been divided. This result has been classified into three levels based on the absolute value of the sensitivity coefficient in the previous steps. The priority allocation rule is: Motors with a high-sensitivity level are allocated the first priority sequence, motors with a medium-sensitivity level are allocated the second priority sequence, and motors with a low-sensitivity level are allocated the third priority sequence. Motors within the same priority sequence are sorted secondarily according to the numerical value of the causal entropy of parameter variation, and the one with a larger numerical value obtains a more forward sub-priority number. Each priority corresponds to a compensation response time window and a resource allocation coefficient. The response time window for the first priority is 5 milliseconds. For example, among two motors with a high-sensitivity level, the causal entropy of parameter variation of motor A is 0.8 and it is allocated the sub-priority 1.1, and the numerical value of motor B is 0.7 and it is allocated the sub-priority 1.2.

[0063] The technical implementation of generating a collaborative compensation strategy by integrating the compensation action instruction, dynamic amplitude limit instruction, and resonance suppression priority instruction of the leading compensation role is as follows: Construct a three-dimensional instruction integration structure. The first dimension is the motor identifier, the second dimension is the instruction type, and the third dimension is the time series. The compensation action instruction of the leading compensation role is written into the leading instruction domain corresponding to the motor identifier, including the compensation current amplitude, phase angle, and waveform generation parameters. The dynamic amplitude limit instruction is written into the amplitude limit constraint domain of all motor identifiers as the hard upper limit of the current output. The resonance suppression priority instruction is converted into a time weight coefficient and written into the priority domain. The calculation formula for the weight coefficient is the reference weight divided by the numerical value of the priority number. The executable compensation parameter set for each motor is generated through cross-domain parsing. For example, the time weight coefficient of the motor with sub-priority 1.1 is 0.91 (reference weight 1.0 / 1.1), and the response time window is set to 5 milliseconds.

[0064] When the leading compensation role executes the compensation action, it gives priority to responding to the dynamic amplitude limit instruction, and the remaining motors execute the compensation action according to the resonance suppression priority. The control logic is as follows: At the start of the compensation cycle, first verify whether the calculated compensation current of the leading compensation role exceeds the dynamic amplitude limit instruction value. If it exceeds, limit the output current of the leading compensation role to the amplitude limit value, and the compensation amount of the exceeded part is defined as the compensation difference to be allocated. The compensation difference to be allocated is distributed to the remaining motors according to the proportion of the resonance suppression priority weight coefficient. Motors with a higher priority bear it first and the distribution ratio is positively correlated with the weight coefficient. The distribution process is iteratively executed until the compensation difference is completely distributed. For example, the leading motor needs to output 90 amperes but the amplitude limit is 80 amperes, resulting in a difference of 10 amperes. The motor with sub-priority 1.1 with a weight coefficient of 0.91 bears 5.5 amperes, and the motor with sub-priority 1.2 with a weight coefficient of 0.83 bears 4.5 amperes.

[0065] Strategy Execution Timing Control: Set a fixed compensation period of 10 milliseconds, which is divided into three sequential execution phases within the period. In the first phase from 0 to 3 milliseconds, the dominant compensation role performs basic compensation. In the second phase from 3 to 6 milliseconds, the high-sensitivity-level motor performs priority compensation. In the third phase from 6 to 10 milliseconds, the medium- and low-sensitivity-level motors perform supplementary compensation. A 1-millisecond overlapping buffer period is set between phases for instruction synchronization.

[0066] Instruction Conflict Resolution Mechanism: When there is a conflict between the dynamic amplitude limit constraint and the priority allocation requirement, a two-level adjustment strategy is initiated. At the first level, the amplitude limit value is adjusted according to the bus voltage deviation. When the voltage deviation exceeds 5%, the amplitude limit relaxation ratio is 5%. At the second level, the response time is compressed according to the resonant energy growth rate. When the growth rate exceeds 10%, the time window is shortened by 30%.

[0067] Compensation Effect Feedback Calibration: After each compensation period ends, the total harmonic distortion rate of the system is collected as a feedback indicator. If the total harmonic distortion rate is higher than the preset safety threshold, the compensation intensity is increased in the order of priority. The increase amplitude is 5% of the reference compensation amount, and at most three levels can be increased each time. After the total harmonic distortion rate reaches the standard, the reference compensation parameters are restored.

[0068] Multi-Ringing Condition Handling: When there are multiple dominant oscillation frequencies in the system, an independent cooperative compensation strategy is generated for each oscillation frequency. The strategy execution adopts a time-slicing rotation mechanism. In each 10-millisecond period, at most three oscillation frequency compensation tasks are processed, and time resources are allocated according to the oscillation energy from high to low.

[0069] Strategy Persistent Storage: The finally generated cooperative compensation strategy is encoded into a structured data message, which includes four fields: motor address code, compensation type code, action parameter set, and timing control code. The data message is stored in a circular strategy buffer, and the buffer capacity is 100 records, which are overwritten and updated in chronological order.

[0070] S6. Reconstruct the output torque command of the speed compensator according to the cooperative compensation strategy and synchronously send it to the corresponding motor for execution. The specific implementation is as follows: The specific implementation process of analyzing the dominant compensation role instruction in the collaborative compensation strategy and determining the compensation torque weight coefficient of the dominant motor is as follows: Extract the dominant compensation role instruction set from the instruction storage area of the collaborative compensation strategy data structure. This instruction set includes the compensation current amplitude, phase angle, and role identification code. The compensation torque weight coefficient is defined as the proportion of the total system compensation torque borne by this motor, and is calculated by dividing the compensation current amplitude of the dominant motor by the system reference compensation current value. The value range of the compensation torque weight coefficient is 60% to 80%, and the specific value is determined by linear interpolation according to the parameter mutation causal entropy value of the dominant motor. The higher the parameter mutation causal entropy value, the greater the compensation torque weight coefficient. For example, when the parameter mutation causal entropy value of the dominant motor is 0.8, the compensation torque weight coefficient obtained by interpolation calculation is 75%.

[0071] The technical implementation of converting the dynamic limit instruction into the torque output upper limit value, and using the upper limit value as the amplitude constraint of all motor torque instructions is as follows: Read the maximum allowable compensation current value recorded in the dynamic limit instruction, and multiply this current value by the real-time torque constant of the motor to obtain the torque output upper limit value. The real-time torque constant is obtained from the motor parameter database, which updates the torque constant calibration value according to the motor temperature and speed in real time. The torque output upper limit value acts as a rigid constraint condition in the generation process of all motor output torque instructions. When the output torque calculated by any motor exceeds this upper limit value, the system automatically corrects it to the upper limit value. For example, when the maximum allowable compensation current is 80 amperes and the real-time torque constant is 0.8 N·m per ampere, the torque output upper limit value is 64 N·m.

[0072] The operation process of allocating the torque response time sequence of each motor according to the resonance suppression priority, and allocating the motor with a higher priority to the earlier execution time slot is as follows: Obtain the resonance suppression priority list, which includes the priority number and sub-priority number of each motor. Establish a time slot allocation mapping rule: The motor sequence with a priority number of 1 is allocated to execute in the time period from 0 to 3 milliseconds after the start of the compensation cycle, the sequence with a priority number of 2 is allocated to execute from 3 to 6 milliseconds, and the sequence with a priority number of 3 is allocated to execute from 6 to 10 milliseconds. The motors within the same priority sequence are arranged in ascending order of the sub-priority number, and each motor is allocated an independent micro-time slot. The length of the micro-time slot is evenly distributed according to the total number of motors in this priority sequence. For example, there are two motors in the priority 1 sequence, and sub-priority 1.1 is allocated 0.0 - 1.5 milliseconds, and sub-priority 1.2 is allocated 1.5 - 3.0 milliseconds.

[0073] The method for reconstructing the output torque command of the speed compensator based on the compensation torque weight coefficient, torque output upper limit value, and torque response timing sequence is as follows: Construct a torque command reconstruction matrix. The row index of the matrix corresponds to the motor identifier, and the column attributes include the compensation torque weight coefficient column, torque output upper limit value column, execution start time column, and duration column. The calculation formula for the output torque command of the main motor is the reference torque command multiplied by the compensation torque weight coefficient, and the minimum value is taken after comparing the calculation result with the torque output upper limit value. The calculation formula for the output torque command of the non-main motor is the reference torque command multiplied by the priority weight coefficient, and the priority weight coefficient is equal to the reference weight coefficient divided by the numerical value of the sub-priority number of the motor. The output torque commands of all motors need to be bound with their timing parameters, including the start time point when the command becomes effective and the duration of continuous execution. For example, for a motor with sub-priority 1.1, the priority weight coefficient is 0.91 (reference weight coefficient 1.0 divided by 1.1), and the reference torque command is 50 N·m, then the output torque command is 45.5 N·m.

[0074] The technical implementation of synchronously transmitting the reconstructed output torque command to each motor controller through the real-time communication bus is as follows: Use the time-triggered Ethernet protocol as the real-time communication bus, and the bus communication cycle is synchronized with the system compensation cycle to be 10 milliseconds. 2 milliseconds before the start moment of the compensation cycle, encapsulate all the output torque commands of the motors that have been reconstructed in the standard data frame format. The standard data frame contains four fields: the target motor address code field, the torque command value field, the command effective timestamp field, and the command duration field. Broadcast and send the data frame through the bus controller to ensure that the transmission delay does not exceed 100 microseconds, and all motor controllers complete the command reception and parsing before the start moment of the compensation cycle. For example, the command encapsulation is completed at 8.0 milliseconds, and it is ensured that all nodes receive successfully before 9.9 milliseconds.

[0075] The control logic for the main motor's output torque command to preferentially apply the torque weight coefficient and the remaining motors to apply it in the order of resonance suppression priority is as follows: At the start moment of the compensation cycle, the main motor immediately executes the output torque command adjusted by the compensation torque weight coefficient. The output torque commands of the non-main motors are stored in the command buffer register of the controller. When the system time reaches the start time point specified by the timing parameters of the motor, the buffer register automatically releases the command to the execution unit. During the execution process, the actual output torque value is monitored in real time. If it is detected that the actual value exceeds the torque output upper limit value, the limiter is immediately activated to clamp the output value to the upper limit value. For example, the main motor executes the torque command at 0.0 milliseconds, the motor with sub-priority 1.1 executes at 0.0 milliseconds, and the motor with sub-priority 1.2 executes at 1.5 milliseconds.

[0076] Timing Synchronization Calibration Mechanism: The IEEE 1588 Precision Time Protocol is used to synchronize the local clocks of each motor controller, and the maximum clock deviation is controlled within 50 microseconds. When the deviation between the actual execution start time and the instruction timestamp is detected to exceed 100 microseconds, a time compensation offset is automatically inserted at the start of the next compensation cycle, and the compensation amount is equal to the deviation value.

[0077] Instruction Pre-Verification Process: The motor controller performs three verifications before executing an instruction: verifying whether the torque instruction value is less than the upper limit of the motor safety torque, verifying whether the instruction effective timestamp is within the current compensation cycle, and verifying whether the target address code is consistent with the local address. If any verification fails, the instruction is discarded and the valid instruction of the previous cycle is enabled, and the fault type is reported through an error code.

[0078] Dynamic Weight Adjustment Strategy: During the execution of the compensation cycle, the change rate of the system resonance energy is monitored in real time. When the change rate exceeds the threshold of 10% per second, the compensation torque weight coefficient of the leading motor is increased by 5%, and at the same time, the priority weight coefficient of the non-leading motor is decreased by 5% proportionally. The adjustment operation takes effect in the next compensation cycle.

[0079] Multi-Cycle Role Rotation: A leading compensation role rotation mechanism is established. The leading motor in the current cycle cannot serve as the leading role again in the next two cycles. When the role is switched, the weight coefficient is reallocated through the total compensation conservation algorithm, and the switching process is completed within 2 milliseconds of the cycle interval. For example, if the weight coefficient of the leading motor in the current cycle is 70%, it will be reduced to 40% after switching, and the difference is transferred to the new leading motor.

[0080] Communication Fault Tolerance: When the bus communication is interrupted for more than 200 microseconds, each motor controller switches to the local compensation mode. The local compensation mode uses fixed parameters: the compensation torque weight coefficient is 50%, the upper limit value of the torque output is 80% of the rated value, and the execution timing uses the default priority. The system continuously monitors the communication status and automatically switches back to the collaborative compensation mode after recovery.

[0081] The entire compensation control process of this embodiment establishes a quantitative causal relationship between parameter mismatch and harmonic disturbance through parameter mutation causal entropy, which is different from the conventional threshold judgment mechanism. Based on the causal strength, the dominant compensation role is dynamically allocated to achieve responsibility focus. The system cooperation vulnerability quantifies the risk of multi-motor coupled oscillation, and its real-time positive correlation mapping with the dynamic amplitude limiting instruction replaces the fixed amplitude limiting mode, synchronously adapting to the changes in system stability. The resonance suppression priority is hierarchically sorted according to the equipment sensitivity level, forming a spatio-temporal gradient response sequence for fault suppression. The above three decision-making elements construct a complementary constraint relationship in the strategy fusion. When the dominant role executes the core amplitude limiting task, the priority sequence dynamically allocates the residual compensation amount, forming a complete compensation demand transfer chain. The torque instruction reconstruction process converts the electrical quantity compensation into mechanical torque regulation, and realizes the precise spatio-temporal decoupling of energy distribution through the binding of weight coefficients and time sequences. The instruction synchronous issuing mechanism ensures the timing consistency of cross-device control, and the role rotation strategy avoids the risk of single-point overload.

[0082] Embodiment 2: Figure 2 A structural schematic diagram of the multi-motor cooperative control system based on speed compensation according to the present invention is given. The multi-motor cooperative control system based on speed compensation includes: A signal acquisition module for real-time acquisition of the operating current signals and terminal voltage signals of each motor in the system; A parameter calculation module for online calculating the real-time resistance parameters and real-time inductance parameters of each motor according to the operating current signals and terminal voltage signals; An entropy network analysis module for calculating the parameter mutation causal entropy between each motor through the transfer entropy algorithm based on the real-time resistance parameters and real-time inductance parameters, and determining the system cooperation vulnerability based on the Granger causal network analysis; An oscillation sensitivity analysis module for extracting the harmonic components of the operating current signal at the detected low-frequency oscillation frequency and analyzing the sensitivity level of each motor parameter mismatch to the current oscillation frequency when detecting that the system has a low-frequency oscillation; A strategy generation module for allocating the dominant compensation role according to the parameter mutation causal entropy, generating a dynamic amplitude limiting instruction according to the system cooperation vulnerability, allocating the resonance suppression priority according to the sensitivity level, and integrating them into a cooperative compensation strategy; An instruction reconstruction module for reconstructing the output torque instruction of the speed compensator according to the cooperative compensation strategy and synchronously issuing it to the corresponding motor for execution.

[0083] The calculations involved in the embodiments are all dimensionless numerical calculations, and the preset parameters and threshold selections in the calculations are set by those skilled in the art according to the actual situation.

[0084] It should be noted that the present invention can be deployed on the device itself to implement embedded applications, or can also run on a PC or other terminal with a user interface, so as to meet various hardware environments and usage requirements.

[0085] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wire (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more collections of available media. The available medium can be a magnetic medium (such as a floppy disk, hard disk, magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0086] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0087] In several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is only a logical function division, and there can be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or modules can be in electrical, mechanical or other forms.

[0088] The module described as a separation component may or may not be physically separated. The component shown as a module may or may not be a physical module. It may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0089] In addition, in each embodiment of the present application, each functional module can be integrated into a processing module, or each module can exist physically alone, or two or more modules can be integrated into one module.

[0090] If the function is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0091] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present application, and all should be covered by the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0092] Finally: The above is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-motor collaborative control method based on speed compensation, characterized in that, Including: S1. Real-time collect the running current signals and terminal voltage signals of each motor in the system; S2. According to the running current signals and terminal voltage signals, online calculate the real-time resistance parameters and real-time inductance parameters of each motor; S3. Based on the real-time resistance parameters and real-time inductance parameters, calculate the parameter mutation causal entropy between each motor through the transfer entropy algorithm, and determine the system collaborative vulnerability based on the Granger causal network analysis; S4. When it is detected that there is a low-frequency oscillation in the system, extract the harmonic components of the running current signal at the detected low-frequency oscillation frequency, and analyze the sensitivity level of each motor parameter mismatch to the current oscillation frequency; S5. Allocate the dominant compensation role according to the parameter mutation causal entropy, generate a dynamic amplitude limiting instruction according to the system collaborative vulnerability, allocate the resonance suppression priority according to the sensitivity level, and integrate them into a collaborative compensation strategy; S6. Reconstruct the output torque instruction of the speed compensator according to the collaborative compensation strategy, and synchronously send it to the corresponding motor for execution.

2. The multi-motor cooperative control method based on speed compensation according to claim 1, wherein Real-time collect the running current signals and terminal voltage signals of each motor in the system, including: Synchronously sample the instantaneous current values of the three-phase windings of each motor to generate running current signals; Synchronously sample the instantaneous values of the input voltages at the driving ends of each motor to generate terminal voltage signals.

3. The multi-motor collaborative control method based on speed compensation according to claim 2, wherein, Among them, the sampling of the running current signals and terminal voltage signals maintains the same time reference and is updated and stored in continuous cycles.

4. The multi-motor collaborative control method based on speed compensation according to claim 1, wherein According to the running current signals and terminal voltage signals, online calculate the real-time resistance parameters and real-time inductance parameters of each motor, including: Establish a parameter calculation model based on the motor stator voltage equation, and the parameter calculation model represents the terminal voltage signal as a function of the running current signal and its derivative; Perform central difference calculation on the running current signals of multiple consecutive sampling periods to obtain current change rate data; Input the terminal voltage signal, running current signal and current change rate data into the parameter calculation model to construct an overdetermined system of equations; Iteratively solve the overdetermined system of equations by the recursive least squares method to update the estimated values of the real-time resistance parameters and real-time inductance parameters; When the relative change amount of the estimated value is less than the convergence threshold for a continuously set number of times, output the real-time resistance parameters and real-time inductance parameters.

5. The multi-motor collaborative control method based on speed compensation according to claim 1, wherein Based on the real-time resistance parameters and real-time inductance parameters, calculate the parameter mutation causal entropy between each motor through the transfer entropy algorithm, and determine the system collaborative vulnerability based on the Granger causal network analysis, including: Extract the real-time resistance parameters and real-time inductance parameters of each motor according to the set time window to form a parameter time series; Calculate the transfer entropy for the parameter time series of every two motors, and the delay order of the transfer entropy is determined according to the minimum point of the mutual information; Normalize the transfer entropy value to the parameter mutation causal entropy, and the parameter mutation causal entropy represents the parameter mutation transfer intensity; Construct a directed weighted network based on the parameter mutation causal entropy between all motors, where the nodes represent motors and the edge weights represent the parameter mutation causal entropy; Perform Granger causality test on the network, and add causal edges when the test statistic exceeds the set significance threshold; Calculate the variance of the node betweenness centrality of the network as the system collaborative vulnerability.

6. The multi-motor collaborative control method based on speed compensation according to claim 1, wherein When low-frequency oscillations are detected in the system, extract the harmonic components of the operating current signal at the detected low-frequency oscillation frequency, and analyze the sensitivity levels of the mismatches of each motor parameter to the current oscillation frequency, including: Identify the low-frequency components in the operating current signal whose amplitudes exceed a set proportion of the fundamental wave amplitude through spectrum analysis. When the frequency of the low-frequency components is within the set low-frequency range, it is determined that low-frequency oscillations exist; When low-frequency oscillations exist, extract the harmonic components with positive and negative set percentage bandwidths centered on the oscillation frequency; Calculate the relative deviation between the measured parameters and the reference parameters of each motor based on the real-time resistance parameters and real-time inductance parameters as the parameter mismatch degree; Establish a linear mapping relationship between the parameter mismatch degree and the harmonic component amplitude, and obtain the sensitivity coefficient through least squares fitting; Divide it into three levels: high-sensitivity level, medium-sensitivity level, and low-sensitivity level according to the absolute value of the sensitivity coefficient; 7. The multi-motor cooperative control method based on speed compensation according to claim 1, wherein Allocate the dominant compensation role according to the causal entropy of parameter variation, generate a dynamic amplitude limiting instruction according to the system collaborative vulnerability, allocate the resonance suppression priority according to the sensitivity level, and integrate them into a collaborative compensation strategy, including: Select motors based on the causal entropy of parameter variation sorted from high to low, and allocate the motor with the largest causal entropy of parameter variation as the dominant compensation role; Read the system collaborative vulnerability, generate a dynamic amplitude limiting instruction according to the system collaborative vulnerability, and the amplitude limiting range of the dynamic amplitude limiting instruction is positively correlated with the system collaborative vulnerability; Allocate the resonance suppression priority from the high-sensitivity level to the low-sensitivity level according to the sensitivity level; Fuse the compensation action instruction, dynamic amplitude limiting instruction, and resonance suppression priority instruction of the dominant compensation role to generate a collaborative compensation strategy; 8. The multi-motor collaborative control method based on speed compensation according to claim 7, characterized in that When the dominant compensation role executes the compensation action, it responds to the dynamic amplitude limiting instruction first, and the other motors execute the compensation action according to the resonance suppression priority; 9. The multi-motor collaborative control method based on speed compensation according to claim 1, characterized in that, Reconstruct the output torque instruction of the speed compensator according to the collaborative compensation strategy and synchronously send it to the corresponding motor for execution, including: Analyze the dominant compensation role instruction in the collaborative compensation strategy to determine the compensation torque weight coefficient of the dominant motor; Convert the dynamic amplitude limiting instruction into the upper limit value of torque output, and the upper limit value is used as the amplitude constraint of all motor torque instructions; Allocate the torque response time sequence of each motor according to the resonance suppression priority, and the motor with a higher priority is allocated a higher execution time slot; Reconstruct the output torque instruction of the speed compensator based on the compensation torque weight coefficient, torque output upper limit value, and torque response time sequence; Synchronously send the reconstructed output torque instruction to each motor controller through the real-time communication bus; The output torque instruction of the dominant motor preferentially applies the torque weight coefficient, and the other motors apply it in the order of resonance suppression priority; 10. A multi-motor collaborative control system based on speed compensation is used to implement the multi-motor collaborative control method based on speed compensation according to any one of claims 1-9, characterized in that, Including: A signal acquisition module for real-time acquisition of the operating current signal and terminal voltage signal of each motor in the system; A parameter calculation module for online calculation of the real-time resistance parameters and real-time inductance parameters of each motor according to the operating current signal and terminal voltage signal; An entropy network analysis module for calculating the causal entropy of parameter variation between each motor through the transfer entropy algorithm based on the real-time resistance parameters and real-time inductance parameters, and determining the system collaborative vulnerability based on the Granger causal network analysis; An oscillation-sensitive analysis module, which is used to extract the harmonic components of the operating current signal at the detected low-frequency oscillation frequency when detecting low-frequency oscillations in the system, and analyze the sensitivity level of each motor parameter mismatch to the current oscillation frequency; A strategy generation module, which is used to assign the dominant compensation role according to the causal entropy of parameter mutation, generate a dynamic amplitude-limiting instruction according to the system collaborative vulnerability, assign the resonance suppression priority according to the sensitivity level, and integrate them into a collaborative compensation strategy; An instruction reconstruction module, which is used to reconstruct the output torque instruction of the speed compensator according to the collaborative compensation strategy and synchronously send it to the corresponding motor for execution.

Images