Servo motor energy-saving control system and method

Abstract

The invention discloses a servo motor energy-saving control system and method, and relates to the technical field of servo motor energy conservation, and the method comprises the steps: obtaining a magnetothermal coupling coefficient based on a magnetic field intensity matrix and a temperature distribution matrix, recognizing a temperature abnormal region, and generating a hot spot mark; on the basis of the magnetothermal coupling coefficient and the hotspot mark, encoding the magnetothermal coupling coefficient into a pulse sequence through a pulse neural network, and generating an energy recovery threshold value and an efficiency threshold value by using a synaptic weight updating mechanism of a hidden layer neuron; and performing dynamic energy recovery on the servo motor based on an energy recovery threshold and an efficiency threshold, collecting a vibration signal after energy recovery, calculating a maximum Lyapunov index, and updating the pulse neural network. According to the method, accurate correlation mapping of magnetic field distortion and local temperature rise is realized through magnetothermal coupling coefficient calculation and hot spot marking, the limitation of traditional independent analysis is broken through, and a physical basis is provided for energy recovery.

Description

Technical Field

[0001] The invention relates to the technical field of servo motor energy saving, and in particular to a servo motor energy saving control system and method. Background Art

[0002] Servo motor energy-saving control technology has received extensive attention in the field of industrial automation in recent years. Traditional methods are mainly based on PID adjustment of current loop and speed loop, combined with vector control to achieve energy efficiency optimization. With the development of sensing technology, the coordinated monitoring of magnetic field strength and temperature distribution has become a research hotspot. Existing technologies can collect motor operating parameters through Hall sensors and thermocouple arrays, and use Kalman filtering, wavelet transform and other methods for data processing. In terms of energy recovery, technologies such as regenerative braking and supercapacitor energy storage have achieved partial energy reuse, and intelligent control algorithms based on neural networks have gradually been applied to motor energy efficiency optimization. However, the existing technology lacks dynamic characterization of the magnetic field-temperature coupling effect, making it difficult to accurately identify the correlation between local overheating areas and energy loss, resulting in a lack of physical basis for setting the energy recovery threshold, which restricts further improvement of energy saving efficiency.

[0003] The existing technology has the following limitations: First, the processing of magnetic field and temperature data is usually carried out independently, lacking quantitative analysis of the magnetothermal coupling effect, and unable to establish a mapping relationship between magnetic field distortion and local temperature rise; second, the setting of energy recovery threshold mostly relies on empirical formulas or static parameters, and does not take into account the dynamic nonlinear characteristics of the motor's operating state, resulting in over-recovery or under-recovery problems. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a servo motor energy-saving control method to solve the problems of lack of correlation between magnetic field and temperature and static energy recovery threshold in the prior art.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a servo motor energy-saving control method, which includes collecting magnetic field strength data and temperature distribution data of the servo motor and preprocessing them, and obtaining a magnetic field strength matrix and a temperature distribution matrix at the same time; based on the magnetic field strength matrix and the temperature distribution matrix, obtaining a magnetothermal coupling coefficient, identifying temperature abnormality areas, and generating hot spot markers; based on the magnetothermal coupling coefficient and the hot spot markers, encoding the magnetothermal coupling coefficient into a pulse sequence through a pulse neural network, and using the synaptic weight update mechanism of hidden layer neurons to generate an energy recovery threshold and an efficiency threshold; based on the energy recovery threshold and the efficiency threshold, dynamically recovering energy from the servo motor, collecting the vibration signal after energy recovery, calculating the maximum Lyapunov exponent, and updating the pulse neural network.

[0007] As a preferred solution of the servo motor energy-saving control method described in the present invention, the preprocessing includes noise filtering, outlier elimination, normalization processing, time-space alignment and data standardization.

[0008] As a preferred solution of the servo motor energy-saving control method of the present invention, the steps of obtaining the magnetic field intensity matrix and the temperature distribution matrix are as follows: Based on the preprocessed magnetic field intensity data, a magnetic field intensity matrix is ​​constructed by grid interpolation method; Based on the preprocessed temperature distribution data, the temperature distribution matrix is ​​constructed by Kriging interpolation method.

[0009] As a preferred solution of the servo motor energy-saving control method of the present invention, the steps of obtaining the magnetic thermal coupling coefficient are as follows: By using collaborative filtering algorithm, the magnetic field intensity eigenvector in the magnetic field intensity matrix and the temperature distribution eigenvector in the temperature distribution matrix are extracted; The singular value decomposition method is used to decompose the magnetic field intensity eigenvector and the temperature distribution eigenvector into eigenvectors, and the magnetothermal coupling coefficient is obtained by combining the tensor product with the nonlinear activation function.

[0010] As a preferred solution of the servo motor energy-saving control method of the present invention, the steps of identifying the abnormal temperature area and generating the hot spot mark are as follows: Based on the magnetothermal coupling coefficient, the temperature abnormality area of ​​the servo motor is identified by the local outlier factor method, and the coupling threshold is calculated by the fuzzy membership function; According to the temperature anomaly area and coupling threshold, hot spot markers are generated through fuzzy logic discrimination method.

[0011] As a preferred solution of the servo motor energy-saving control method of the present invention, the steps are as follows: based on the magnetothermal coupling coefficient and the hot spot mark, the magnetothermal coupling coefficient is encoded into a pulse sequence through a pulse neural network, and the synaptic weight update mechanism of the hidden layer neurons is used to generate the energy recovery threshold and the efficiency threshold. Based on the hot spot marking, the magnetothermal coupling coefficient and are converted into a pulse sequence through a pulse encoder; The pulse sequence is input into the hidden layer neurons, and each neuron updates the membrane potential of each neuron according to the input pulse sequence; Define the neuron's membrane potential threshold based on the historical spike firing rate and membrane potential dynamic range; Calculate the pulse firing frequency based on the updated neuron's membrane potential; By analyzing historical energy recovery data, the linear regression method is used to fit the relationship between pulse emission frequency and energy recovery value, and the energy recovery function is constructed using a smoothing factor. By analyzing historical efficiency data, a nonlinear fitting method is used to capture the nonlinear effect of pulse emission frequency on servo motor efficiency, and an efficiency function is constructed in combination with the saturation factor. An energy recovery threshold and an efficiency threshold are generated according to the pulse emission frequency, the energy recovery function and the efficiency function.

[0012] As a preferred solution of the servo motor energy-saving control method of the present invention, the steps are as follows: based on the energy recovery threshold and the efficiency threshold, the servo motor is dynamically energy recovered, and the vibration signal after energy recovery is collected, the maximum Lyapunov exponent is calculated, and the pulse neural network is updated. Real-time monitoring of the real-time energy parameters of the servo motor. When the real-time energy parameters exceed the energy recovery threshold, energy recovery is performed, and the intensity of energy recovery is adjusted according to the efficiency threshold. Based on the vibration signal of the servo motor, the maximum Lyapunov exponent is calculated by phase space reconstruction and small data method. Update the spiking neural network according to the maximum Lyapunov exponent.

[0013] In a second aspect, the present invention provides a servo motor energy-saving control system, including a data acquisition module, a hotspot marking module, a threshold generation module and an energy recovery module; the data acquisition module is used to collect the magnetic field strength data and temperature distribution data of the servo motor and perform preprocessing, and simultaneously obtain the magnetic field strength matrix and the temperature distribution matrix; the hotspot marking module is used to obtain the magnetothermal coupling coefficient based on the magnetic field strength matrix and the temperature distribution matrix, and identify the temperature abnormality area to generate the hotspot marking; the threshold generation module is used to encode the magnetothermal coupling coefficient into a pulse sequence through a pulse neural network based on the magnetothermal coupling coefficient and the hotspot marking, and generate an energy recovery threshold and an efficiency threshold by using the synaptic weight update mechanism of the hidden layer neurons; the energy recovery module is used to perform dynamic energy recovery on the servo motor based on the energy recovery threshold and the efficiency threshold, and collect the vibration signal after energy recovery, calculate the maximum Lyapunov exponent, and update the pulse neural network.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the servo motor energy-saving control method as described in the first aspect of the present invention is implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the servo motor energy-saving control method as described in the first aspect of the present invention is implemented.

[0016] The beneficial effects of the present invention are as follows: through the calculation of the magnetothermal coupling coefficient and the marking of hot spots, accurate correlation mapping between magnetic field distortion and local temperature rise is achieved, breaking through the limitations of traditional independent analysis and providing a physical basis for energy recovery; the magnetothermal coupling coefficient is encoded into a pulse sequence through a pulse neural network, and the synaptic weight update mechanism of the hidden layer neurons is utilized to improve the real-time response capability of the servo motor to energy recovery and efficiency optimization during operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0018] Figure 1 This is a flow chart of the servo motor energy-saving control method in Example 1.

[0019] Figure 2 Schematic diagram of the servo motor energy-saving control system in Example 1.

[0020] Figure 3 This is a flow chart for generating hotspot markers in Example 1.

[0021] Figure 4 This is a flow chart of the operation of the pulse neural network in Example 1. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0025] Example 1, reference Figure 1~Figure 4 , this embodiment provides a servo motor energy-saving control method, comprising the following steps: S1, collecting the magnetic field strength data and temperature distribution data of the servo motor and preprocessing them, and obtaining the magnetic field strength matrix and temperature distribution matrix at the same time; The magnetic field strength data of the servo motor is collected by the Hall sensor, and the temperature distribution data is collected by the thermocouple array.

[0026] Preprocessing includes noise filtering, outlier removal, normalization, spatiotemporal alignment, and data standardization of magnetic field intensity data and temperature distribution data.

[0027] Noise filtering uses a low-pass filter to process the magnetic field intensity data collected by the Hall sensor and the temperature distribution data collected by the thermocouple array to remove high-frequency interference signals and retain valid data components.

[0028] Outlier elimination The magnetic field intensity data and temperature distribution data were screened using the 3σ criterion, and data points that were beyond three times the standard deviation of the mean were eliminated to ensure data reliability.

[0029] The normalization process linearly scales the magnetic field intensity data based on the preset magnetic saturation threshold (0.5T-1.8T), and standardizes the temperature distribution data according to the ambient temperature reference value (25°C) and the maximum allowable temperature rise (150°C), so that the data are unified to the same dimension.

[0030] The time-space alignment synchronizes the magnetic field intensity data and the temperature distribution data through timestamp matching, ensuring that the magnetic field intensity value and the temperature value at the same time point correspond and eliminating time deviation.

[0031] Data standardization converts the processed magnetic field intensity data and temperature distribution data into a 4×3 dimensional magnetic field intensity matrix and temperature distribution matrix that can be directly used for analysis, completing the preprocessing process.

[0032] Based on the preprocessed magnetic field intensity data, the magnetic field intensity matrix is ​​constructed by grid interpolation method.

[0033] Furthermore, the preprocessed magnetic field strength data contains discrete sampling points at different spatial positions. The grid interpolation method first divides the spatial area of ​​the servo motor into a regular grid of 4 rows and 3 columns. Then, based on the magnetic field strength values ​​of the sampling points, the bilinear interpolation algorithm is used to calculate the magnetic field strength values ​​of the unsampled points in the grid to ensure that each grid cell is filled with magnetic field strength data. Finally, the magnetic field strength values ​​of all grid cells are arranged by position to generate a 4×3-dimensional magnetic field strength matrix.

[0034] Based on the preprocessed temperature distribution data, the temperature distribution matrix is ​​constructed by Kriging interpolation method.

[0035] Furthermore, spatial correlation analysis is first performed, and the spatial autocorrelation analysis method commonly used in geographic information systems (GIS) is used to calculate the spatial distance matrix and temperature difference matrix between each sampling point. The global spatial autocorrelation is evaluated by calculating Moran's I, and the local spatial association index (LISA) is used to identify local aggregation patterns. This step requires setting a reasonable neighborhood search radius, and the adaptive bandwidth determination method is usually used. Then the temperature value is estimated, and the ordinary Kriging method in spatial interpolation technology is used. When establishing the spatial weight matrix, the inverse distance weighted method is used to determine the influence weight of the neighboring points. For each point to be estimated, the optimal number of neighboring sampling points is selected to participate in the calculation, and the Akaike Information Criterion (AIC) is usually used to determine the optimal number of neighborhood points. The Kriging equations need to be solved during the calculation process, and the LU decomposition method is used for matrix inversion. Finally, gridding is performed to arrange the estimated temperature values ​​according to a regular grid. The grid division adopts the equal spacing division method, which is evenly divided into 3 columns in the horizontal direction and 4 rows in the vertical direction. The temperature value of each grid cell is determined according to the coordinates of its center point, and the nearest neighbor method is used for coordinate matching. The final preprocessed temperature distribution data contains discrete sampling points at different spatial locations. The Kriging interpolation method first analyzes the spatial correlation between the sampling points through GIS. Then, based on the spatial correlation between the sampling points, the Akaike information criterion is used to determine the optimal number of neighborhood points. The number of points is estimated based on the optimal neighborhood, and the temperature value of the unsampled point is estimated. Finally, the temperature estimation values ​​of all grid cells are arranged according to a regular grid. The grid division adopts the equal spacing division method, which is evenly divided into 3 columns in the horizontal direction and 4 rows in the vertical direction. The temperature value of each grid cell is determined according to the coordinates of its center point, and the nearest neighbor method is used for coordinate matching. A 4×3-dimensional temperature distribution matrix is ​​generated.

[0036] S2. Based on the magnetic field intensity matrix and the temperature distribution matrix, the magnetic thermal coupling coefficient is obtained, and the temperature abnormality area is identified to generate a hot spot mark; By using collaborative filtering algorithm, the magnetic field intensity eigenvector in the magnetic field intensity matrix and the temperature distribution eigenvector in the temperature distribution matrix are extracted; It should be noted that the principal component analysis (PCA) was first used to reduce the dimension of the magnetic field intensity matrix, and the principal components with a cumulative contribution rate of more than 95% were retained as the magnetic field eigenvectors; at the same time, the non-negative matrix factorization (NMF) technology was used to process the temperature distribution matrix, and the temperature eigenvector was obtained by iterative decomposition using the alternating least squares (ALS) method. The maximum information coefficient (MIC) was then used to evaluate the correlation between the two sets of eigenvectors, and the eigenpairs with a correlation coefficient greater than 0.8 were selected.

[0037] The singular value decomposition method is used to decompose the magnetic field intensity eigenvector and the temperature distribution eigenvector into eigenvectors to capture the implicit relationship between the two. The magnetic thermal coupling coefficient is obtained by combining the tensor product with the nonlinear activation function. The expression is: ; in, represents the magnetothermal coupling coefficient, represents the first The magnetic field strength eigenvector, represents the temperature distribution matrix The temperature distribution feature vector, represents the tensor product operator, represents the weight matrix for adjusting the tensor product operation of the magnetic field intensity eigenvector and the temperature eigenvector, represents the total number of magnetic field strength vectors, Represents the total number of temperature distribution eigenvectors; It should be noted that the magnetic field intensity distribution matrix is ​​first decomposed to extract Orthogonal magnetic field strength eigenvectors ; At the same time, the temperature distribution matrix is ​​decomposed to extract Orthogonal temperature distribution eigenvectors ; Then, according to the expression requirements, all possible combinations of eigenvectors (a total of × Perform a tensor product operation on , generate high-order interaction features; then each tensor product result is combined with the learnable weight matrix Perform a dot multiplication and for all and The summation result of Apply ReLU nonlinear activation function; the final output scalar value This is the desired magnetothermal coupling coefficient.

[0038] Based on the magnetothermal coupling coefficient, the temperature abnormality area of ​​the servo motor is identified by the local outlier factor method; It should be noted that the magnetothermal coupling coefficient is first correlated with the temperature value in the temperature distribution matrix to generate comprehensive characteristic data reflecting the relationship between temperature distribution and magnetic field strength. Then the local outlier factor method is used to calculate the comprehensive characteristic data to evaluate the degree of difference between each temperature area and the surrounding area. For temperature areas with a significantly higher degree of difference than the surrounding area, they are marked as temperature abnormal areas. At the same time, combined with the actual temperature value in the temperature distribution matrix, the abnormal characteristics of the marked area are further verified. Finally, the temperature abnormal area of ​​the servo motor is output.

[0039] Further, the specific process of association analysis: The correlation analysis was performed using the Pearson Correlation Coefficient method. First, the magnetothermal coupling coefficient matrix and the temperature distribution matrix were spatially aligned to ensure that the magnetic field characteristics of each grid unit corresponded to the temperature value one by one. The correlation coefficient matrix was obtained by calculating the ratio of the covariance and standard deviation of the magnetothermal coupling coefficient and the temperature value at each spatial position. The analysis results showed that the two showed a significant positive correlation under normal operating conditions (for example, >0.7, while outlier regions show weakened correlations (e.g. <0.3) or negative correlation characteristics. The correlation coefficient matrix is ​​the comprehensive characteristic data for subsequent analysis.

[0040] Local outlier factor calculation process: The k-nearest neighbors algorithm is used to calculate the local outlier factor (LOF). Taking each temperature region as the center, find its The nearest neighbor region is calculated. The local reachable density ratio of the region relative to the neighborhood is calculated, and this ratio is the LOF value. In the specific calculation, the k-distance neighborhood of each point is first determined, and then the local reachable density (LRD) is calculated. Finally, the LOF value is obtained by comparing the LRD ratio of the target point and the neighborhood.

[0041] Difference assessment criteria: The degree of difference is quantitatively evaluated through the statistical distribution of LOF values. The difference threshold K1 is set. When the LOF value exceeds the mean K1 of the same layer area, it is judged to be a significant difference. This difference is manifested as: 1) the spatial continuity is destroyed, and the magnetic and thermal correlation in the abnormal area suddenly decreases; 2) it indicates a sudden change in the temperature gradient.

[0042] Based on the magnetothermal coupling coefficient, the coupling threshold is calculated through the fuzzy membership function, and the expression is: ; in, is the coupling threshold, is the magnetothermal coupling coefficient The degree of membership to the hotspot ranges from [0, 1]. represents the slope parameter of the fuzzy membership function, represents the offset parameter of the fuzzy membership function; It should be noted that the magnetothermal coupling coefficient is first input into the fuzzy membership function, and its membership value is calculated according to the degree of association between the magnetothermal coupling coefficient and the hotspot area. The membership value reflects the possibility that the magnetothermal coupling coefficient belongs to the hotspot area, ranging from zero to one. Then, according to the characteristics of the membership function, the critical point at which the magnetothermal coupling coefficient reaches the hotspot area is determined, that is, the coupling threshold. The coupling threshold is used to determine whether the magnetothermal coupling coefficient belongs to the hotspot area, providing a basis for subsequent hotspot marking.

[0043] According to the temperature anomaly area and coupling threshold, hot spot markers are generated through fuzzy logic discrimination method.

[0044] Furthermore, we first analyze whether the temperature value of each unit in the temperature distribution matrix is ​​located in the temperature anomaly area. If it is located in the abnormal area, we further check whether the magnetic thermal coupling coefficient of the unit exceeds the coupling threshold calculated by the fuzzy membership function; if both conditions are met, the unit is marked as a hotspot. For example, if the temperature value of a unit is located in the abnormal area and the magnetic thermal coupling coefficient exceeds the threshold, the hotspot is marked as 1, otherwise it is marked as 0.

[0045] S3, based on the magnetothermal coupling coefficient and hotspot marking, the magnetothermal coupling coefficient is encoded into a pulse sequence through a pulse neural network, and the synaptic weight update mechanism of the hidden layer neurons is used to generate the energy recovery threshold and efficiency threshold; Based on the hot spot mark, the magnetothermal coupling coefficient is calculated by the pulse encoder. and converted into pulse trains; It should be noted that the pulse encoder first receives the magnetothermal coupling coefficient as an input signal, and performs feature weighting processing in combination with the hotspot mark. The frequency and intensity of pulse emission are determined according to the numerical characteristics of the magnetothermal coupling coefficient and the hotspot mark status. The pulse encoder adopts an encoding rule based on amplitude-frequency conversion to convert continuous magnetothermal coupling coefficient values ​​into discrete pulse signals. During the conversion process, the numerical value of the magnetothermal coupling coefficient and the hotspot mark jointly determine the density of pulse emission. For areas marked as hotspots, the corresponding magnetothermal coupling coefficient will produce a higher frequency pulse sequence, while the areas without hotspots will maintain the conventional encoding method. The final output pulse sequence completely retains the characteristic information of the original magnetothermal coupling coefficient, and reflects the spatial characteristics of the hotspot mark through the difference in pulse frequency, providing a standardized signal form for the input of the subsequent pulse neural network.

[0046] The pulse sequence is input into the hidden layer neurons. Each neuron updates the membrane potential of each neuron according to the input pulse sequence. The expression is: ; in, Indicates the current time Updated The membrane potential of the neuron, Indicates the previous time Time The membrane potential of a neuron, Indicates The synaptic weight for each input spike is Indicates at time Whether to issue a pulse command value when is the total number of input pulses, It is The precise moment at which an input pulse actually arrives at a neuron's synapse; It should be noted that, first, the hidden layer neurons receive input signals from the pulse sequence, and each input pulse affects the membrane potential of the neuron. The neuron adjusts the degree of change of its membrane potential according to the synaptic weight of the input pulse. The pulse with a larger weight has a greater impact on the membrane potential. At the same time, the neuron determines whether the membrane potential needs to be updated at the current moment based on the time characteristics of the input pulse. If the input pulse arrives at the neuron synapse at a specific moment (the specific moment refers to the exact time point when the input pulse signal actually arrives at the neuron synapse ) arrives, the neuron will cumulatively update the membrane potential based on the weight and arrival time of the pulse. Ultimately, the membrane potential of the neuron reflects the combined impact of all input pulses.

[0047] Define the neuron's membrane potential threshold based on historical spike firing rate and membrane potential dynamic range ; It should be noted that the dynamic range of membrane potential refers to the range of changes in neuronal membrane potential within a specific time, reflecting the degree of activity of the neuron in response to input pulses. The upper limit of the dynamic range of membrane potential is the maximum value of the neuron membrane potential, which indicates the potential level of the neuron when it is stimulated by a strong input pulse; the lower limit is the minimum value of the neuron membrane potential, which indicates the potential level of the neuron when it is not stimulated by an input pulse or the input pulse is weak. By analyzing the dynamic range of membrane potential, we can understand the sensitivity and responsiveness of neurons to input pulses, thereby providing a basis for defining the membrane potential threshold. The size of the membrane potential dynamic range directly affects the setting of the membrane potential threshold. A larger dynamic range usually corresponds to a higher membrane potential threshold, and a smaller dynamic range corresponds to a lower membrane potential threshold.

[0048] Based on the updated neuron's membrane potential Calculate the pulse frequency , the expression is: ; By analyzing historical energy recovery data, the linear regression method is used to fit the relationship between pulse emission frequency and energy recovery value, and the energy recovery function is constructed using a smoothing factor. , the expression is: ; in, is the fitting coefficient, which reflects the influence of pulse emission frequency on energy recovery. is the offset; It should be noted that the fitting coefficient The acquisition process is as follows: First, collect the pulse emission frequency data of the servo motor under different working conditions, and record the corresponding energy recovery efficiency. The relationship between the pulse emission frequency and the energy recovery efficiency is fitted by the least squares method to obtain a linear regression equation reflecting the correlation between the two. The slope of the equation is the fitting coefficient , and its numerical value directly reflects the influence of pulse emission frequency on energy recovery efficiency.

[0049] Offset The process of obtaining is as follows: After determining the fitting coefficient Based on this, the energy recovery efficiency benchmark value when the pulse emission frequency is zero is analyzed. The energy recovery efficiency of the servo motor without pulse input is measured experimentally, combined with the fitting coefficient The calculation result of the regression equation is used to determine the intercept value, which is the offset. .Offset It reflects the fundamental influence of other factors besides pulse emission frequency on energy recovery efficiency.

[0050] Other factors specifically refer to changes in ambient temperature during servo motor operation, friction losses in mechanical transmission components, fluctuations in electromagnetic field strength, supply voltage stability, instantaneous changes in mechanical loads, thermal losses in winding resistance, demagnetization effects of permanent magnets, bearing lubrication conditions, and efficiency characteristics of power electronic converters.

[0051] By analyzing historical efficiency data, a nonlinear fitting method is used to capture the nonlinear effect of pulse emission frequency on servo motor efficiency, and an efficiency function is constructed in combination with the saturation factor. , the expression is: ; in, is the saturation factor, the stable value of the control function, is a nonlinear coefficient used to control the rate at which the pulse emission frequency affects the efficiency. is the base of natural logarithms; It should be noted that the historical efficiency data of the servo motor at different pulse emission frequencies are first collected, and the variation between the pulse emission frequency and the efficiency is analyzed by the nonlinear fitting method. The key characteristic parameters that affect the change of efficiency are determined based on the fitting results, including the saturation factor of the stable value of the control function and the nonlinear coefficient of the control influence rate. These two characteristic parameters are substituted into a mathematical relationship with asymptotic characteristics to form a function that can accurately describe the effect of the pulse emission frequency on the efficiency.

[0052] The process of determining the saturation factor and the nonlinear coefficient is as follows: First, a database of the corresponding relationship between the pulse emission frequency and the energy recovery efficiency is established through the measured data of the servo motor under various working conditions. A nonlinear regression analysis is performed on the measured data in the database, and the least squares method is used to fit the curve to find the mathematical model that best reflects the relationship between the pulse emission frequency and efficiency. During the fitting process, the model parameters are adjusted through an iterative optimization algorithm to minimize the error between the predicted value and the measured value. In the final converged optimization result, the parameter that controls the stable value of the curve is the saturation factor, and the parameter that determines the rate of change of the curve is the nonlinear coefficient.

[0053] Based on the pulse frequency and energy recovery function , calculate the energy recovery threshold, the expression is: ; in, is the energy recovery threshold, Indicates at time The energy recovery value at this time is the pulse emission frequency. Mapped to energy recovery value, Indicates at time Whether to issue a pulse command value when Indicates the current time. Indicates the specific time point of pulse emission; It should be noted that first, a complete record of the pulse emission frequency changing over time is obtained, and the pulse emission frequency at each moment is input into the energy recovery function to obtain the corresponding energy recovery value. Then the specific time points of all pulse emission are identified, and the values ​​output by the energy recovery function are extracted at these time points. Finally, the energy recovery values ​​at these time points are accumulated and summed to obtain the final energy recovery threshold. This threshold reflects the total amount of energy recovery that can be achieved at the critical pulse emission moment.

[0054] Based on the pulse frequency and energy recovery function , calculate the efficiency threshold, the expression is: ; in, is the efficiency threshold, Indicates at time The efficiency value when .

[0055] It should be noted that first, the timing data of the pulse emission frequency is collected, the frequency value of each sampling point is input into the efficiency function, and the efficiency value corresponding to each moment is calculated. Then the actual emission time point of the pulse signal is detected, and the output result of the efficiency function is extracted at these specific moments to form an efficiency threshold. This efficiency threshold represents the overall efficiency level that can be achieved when the pulse event occurs.

[0056] S4. Based on the energy recovery threshold and efficiency threshold, the servo motor is dynamically energy recovered, and the vibration signal after energy recovery is collected, the maximum Lyapunov exponent is calculated, and the pulse neural network is updated.

[0057] Real-time monitoring of the real-time energy parameters of the servo motor. When the real-time energy parameters exceed the energy recovery threshold When the energy is recovered, the efficiency threshold Adjust the intensity of energy recovery, the expression is: ; in, Indicates at time Energy recovery power at Indicates at time Whether to issue a pulse command value when Indicates the real-time power of the servo motor, Indicates the start time of energy recovery; It should be noted that the real-time energy parameters of the servo motor include real-time feedback power and real-time output power.

[0058] During the energy recovery process, the vibration signal of the servo motor is collected through the vibration sensor; Based on the vibration signal of the servo motor, the maximum Lyapunov exponent is calculated by phase space reconstruction and small data method, and the expression is: ; in, is the maximum Lyapunov exponent, It's in time The phase space trajectory reconstructed based on the vibration signal is It's in time The phase space trajectory reconstructed based on the vibration signal is It's in time The phase space trajectory reconstructed based on the vibration signal is The time interval representing the evolution time of adjacent trajectories.

[0059] It should be noted that, first, the data of the vibration signal changing with time is collected, then the vibration signal is delayed at a fixed time interval to generate multiple time series, and the time series are combined into a multidimensional vector, that is, a trajectory point in the phase space. By repeating this process, a complete phase space trajectory is formed.

[0060] According to the maximum Lyapunov exponent , update the spiking neural network.

[0061] It should be noted that the current operating status of the equipment is first identified based on the maximum Lyapunov exponent. When the maximum Lyapunov exponent shows chaotic characteristics, the learning ability of the pulse neural network is immediately improved, and the adjustment process of the neuron connection weights is accelerated; when the maximum Lyapunov exponent shows stable characteristics, the existing learning intensity is maintained. Subsequently, the plasticity regulation mechanism based on pulse timing is started to accurately regulate the connection relationship of all neurons, focusing on strengthening the neuron connection with clear timing association, while appropriately weakening the connection with insignificant timing association. This regulation process fully integrates the real-time vibration characteristics and long-term evolution laws, and through continuous optimization, ensures that the pulse neural network parameters are optimally matched with the equipment operating status. Finally, the update of the pulse neural network is completed.

[0062] The present embodiment also provides a servo motor energy-saving control system, including: a data acquisition module, a hotspot marking module, a threshold generation module and an energy recovery module; the data acquisition module is used to collect the magnetic field strength data and temperature distribution data of the servo motor and perform preprocessing, and simultaneously obtain the magnetic field strength matrix and the temperature distribution matrix; the hotspot marking module is used to obtain the magnetothermal coupling coefficient based on the magnetic field strength matrix and the temperature distribution matrix, and identify the temperature abnormality area to generate the hotspot marking; the threshold generation module is used to encode the magnetothermal coupling coefficient into a pulse sequence through a pulse neural network based on the magnetothermal coupling coefficient and the hotspot marking, and generate an energy recovery threshold and an efficiency threshold by using the synaptic weight update mechanism of the hidden layer neurons; the energy recovery module is used to perform dynamic energy recovery on the servo motor based on the energy recovery threshold and the efficiency threshold, and collect the vibration signal after energy recovery, calculate the maximum Lyapunov exponent, and update the pulse neural network.

[0063] This embodiment also provides a computer device, which is suitable for the servo motor energy-saving control method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the servo motor energy-saving control method proposed in the above embodiment.

[0064] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0065] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the servo motor energy-saving control method proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0066] In summary, the present invention achieves accurate correlation mapping between magnetic field distortion and local temperature rise through: magnetothermal coupling coefficient calculation and hotspot marking, breaking through the limitations of traditional independent analysis and providing a physical basis for energy recovery; encoding the magnetothermal coupling coefficient into a pulse sequence through a pulse neural network, and utilizing the synaptic weight update mechanism of hidden layer neurons to improve the real-time response capability of the servo motor to energy recovery and efficiency optimization during operation.

[0067] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A servo motor energy-saving control method, characterized in that: include, Collect the magnetic field strength data and temperature distribution data of the servo motor and perform preprocessing, and simultaneously obtain the magnetic field strength matrix and temperature distribution matrix; Based on the magnetic field intensity matrix and temperature distribution matrix, the magnetic thermal coupling coefficient is obtained, and the temperature abnormality area is identified to generate hot spot markers; Based on the magnetothermal coupling coefficient and hotspot markers, the magnetothermal coupling coefficient is encoded into a pulse sequence through a spiking neural network, and the energy recovery threshold and efficiency threshold are generated by using the synaptic weight update mechanism of the hidden layer neurons; Based on the energy recovery threshold and efficiency threshold, dynamic energy recovery is performed on the servo motor, and the vibration signal after energy recovery is collected, the maximum Lyapunov exponent is calculated, and the pulse neural network is updated.

2. The servo motor energy-saving control method according to claim 1, characterized in that: The preprocessing includes noise filtering, outlier removal, normalization, spatiotemporal alignment and data standardization.

3. The servo motor energy-saving control method according to claim 1, characterized in that: The steps of obtaining the magnetic field intensity matrix and the temperature distribution matrix are as follows: Based on the preprocessed magnetic field intensity data, a magnetic field intensity matrix is ​​constructed by grid interpolation method; Based on the preprocessed temperature distribution data, the temperature distribution matrix is ​​constructed by Kriging interpolation method.

4. The servo motor energy-saving control method according to claim 3, characterized in that: The steps of obtaining the magnetothermal coupling coefficient are as follows: By using collaborative filtering algorithm, the magnetic field intensity eigenvector in the magnetic field intensity matrix and the temperature distribution eigenvector in the temperature distribution matrix are extracted; The singular value decomposition method is used to decompose the magnetic field intensity eigenvector and the temperature distribution eigenvector into eigenvectors, and the magnetothermal coupling coefficient is obtained by combining the tensor product with the nonlinear activation function.

5. The servo motor energy-saving control method according to claim 4, characterized in that: The steps of identifying the abnormal temperature area and generating the hot spot mark are as follows: Based on the magnetothermal coupling coefficient, the temperature abnormality area of ​​the servo motor is identified by the local outlier factor method, and the coupling threshold is calculated by the fuzzy membership function; According to the temperature anomaly area and coupling threshold, hot spot markers are generated through fuzzy logic discrimination method.

6. The servo motor energy-saving control method according to claim 1, characterized in that: Based on the magnetothermal coupling coefficient and hotspot marking, the magnetothermal coupling coefficient is encoded into a pulse sequence through a pulse neural network, and the synaptic weight update mechanism of the hidden layer neurons is used to generate the energy recovery threshold and the efficiency threshold. The steps are as follows: Based on the hot spot marking, the magnetothermal coupling coefficient and are converted into a pulse sequence through a pulse encoder; The pulse sequence is input into the hidden layer neurons, and each neuron updates the membrane potential of each neuron according to the input pulse sequence; Define the neuron's membrane potential threshold based on the historical spike firing rate and membrane potential dynamic range; Calculate the pulse firing frequency based on the updated membrane potential of the neuron; By analyzing historical energy recovery data, the linear regression method is used to fit the relationship between pulse emission frequency and energy recovery value, and the energy recovery function is constructed using a smoothing factor. By analyzing historical efficiency data, a nonlinear fitting method is used to capture the nonlinear effect of pulse emission frequency on servo motor efficiency, and an efficiency function is constructed in combination with the saturation factor. An energy recovery threshold and an efficiency threshold are generated according to the pulse emission frequency, the energy recovery function and the efficiency function.

7. The servo motor energy-saving control method according to claim 6, characterized in that: Based on the energy recovery threshold and the efficiency threshold, the servo motor is dynamically energy recovered, and the vibration signal after energy recovery is collected, the maximum Lyapunov exponent is calculated, and the pulse neural network is updated. The steps are as follows: Real-time monitoring of the real-time energy parameters of the servo motor. When the real-time energy parameters exceed the energy recovery threshold, energy recovery is performed, and the intensity of energy recovery is adjusted according to the efficiency threshold. Based on the vibration signal of the servo motor, the maximum Lyapunov exponent is calculated by phase space reconstruction and small data method. Update the spiking neural network according to the maximum Lyapunov exponent.

8. A servo motor energy-saving control system, based on the servo motor energy-saving control method according to any one of claims 1 to 7, characterized in that: Including, data acquisition module, hot spot marking module, threshold generation module and energy recovery module; A data acquisition module is used to collect the magnetic field strength data and temperature distribution data of the servo motor and perform preprocessing, and simultaneously obtain the magnetic field strength matrix and the temperature distribution matrix; A hotspot marking module is used to obtain the magnetic thermal coupling coefficient based on the magnetic field intensity matrix and the temperature distribution matrix, identify the temperature abnormality area, and generate hotspot markings; A threshold generation module is used to encode the magnetothermal coupling coefficient into a pulse sequence through a pulse neural network based on the magnetothermal coupling coefficient and the hot spot mark, and generate an energy recovery threshold and an efficiency threshold by using a synaptic weight update mechanism of hidden layer neurons; The energy recovery module is used to dynamically recover the energy of the servo motor based on the energy recovery threshold and the efficiency threshold, collect the vibration signal after energy recovery, calculate the maximum Lyapunov exponent, and update the pulse neural network.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the servo motor energy-saving control method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the servo motor energy-saving control method described in any one of claims 1 to 7 are implemented.

Images