A servo steering engine adaptive parameter adjustment method and system and storage medium

Abstract

The application relates to the technical field of servo control, and discloses a servo steering engine adaptive parameter adjustment method and system and a storage medium. The method comprises the following steps: fusing environment data into a comprehensive environment vector to generate a smoothed environment feature set; calculating a deviation vector to determine a significant feature in an environment change period; if the significant feature exceeds a preset threshold, extracting specific data from the environment feature set to calculate a parameter change rate range; mapping a parameter adjustment sequence in combination with historical operation data to calculate a parameter mapping error; if the parameter mapping error exceeds a preset tolerance, iteratively updating an adaptive adjustment coefficient to obtain an optimized adjustment coefficient set; verifying parameter consistency and boundary compliance under a constraint condition, adjusting parameter combination weight distribution and optimizing to obtain a final control parameter combination; and analyzing a stability index to judge system adaptability and generate an operation state evaluation result. The application improves the precision of servo steering engine adaptive parameter adjustment.

Description

Technical Field

[0001] This application relates to the field of servo control technology, and in particular to a method, system and storage medium for adaptive parameter adjustment of a servo motor. Background Technology

[0002] As the core actuator of precision equipment, the control accuracy and operational stability of servo motors directly determine the reliability and performance of the entire system. In practical applications, servo motors often face continuous environmental interference such as temperature fluctuations, humidity changes, and unstable power supply voltage. They are also susceptible to sudden external shocks such as airflow impacts and mechanical vibrations. These factors interact to form complex interference, causing the servo motor control parameters to gradually deviate from the design values, resulting in problems such as output response delay and jitter.

[0003] Existing servo motor control methods mostly employ fixed parameters or simple feedback adjustment modes, lacking the ability to specifically handle complex and compound disturbances. When faced with impact disturbances, the raw environmental data contains a large amount of noise, and directly using it for parameter adjustment can easily lead to over-correction or under-correction, causing the system to remain in an unstable state for a long time. In addition, existing methods cannot effectively separate abnormal fluctuations caused by impacts from normal environmental changes, making it difficult to accurately identify the intensity, duration, and energy distribution characteristics of disturbances. Parameter adjustments lack specificity, ultimately causing the servo motor to accumulate deviations under repeated disturbances, resulting in a continuous deterioration in response stability. Therefore, there is an urgent need for an adaptive adjustment method that can effectively filter noise, extract impact characteristics, and dynamically optimize parameters to solve the technical challenges of insufficient control accuracy and stability of servo motors under complex disturbances. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a servo motor adaptive parameter adjustment method, system, and storage medium, which enables dynamic optimization and precise control of servo motor systems subjected to multiple factors in complex environments.

[0005] In a first aspect, this application provides a servo motor adaptive parameter adjustment method, the method comprising:

[0006] The environmental data of the servo motor system during operation is collected by multiple sensors, the environmental data is fused into a comprehensive environmental vector, the comprehensive environmental vector is filtered for noise, and a smoothed environmental feature set is generated.

[0007] Based on the environmental feature set, the deviation vector between the current working condition and the preset benchmark working condition is calculated. Based on the deviation vector, the significant features within the environmental change period are determined. If the significant features exceed the preset threshold, specific data is extracted from the environmental feature set, and the parameter change rate range is calculated based on the specific data.

[0008] Based on the range of parameter change rates, obtain the parameter adjustment sequence under various working conditions, map the parameter adjustment sequence with historical operating data, calculate the parameter mapping error between the new working condition parameter set and the actual working condition, and if the parameter mapping error exceeds the preset tolerance, iteratively update the adaptive adjustment coefficient to obtain the optimized adjustment coefficient set.

[0009] By integrating the constraints in the set of adjustment coefficients, the consistency of parameters and boundary compliance under the constraints are verified through simulation technology. The weight distribution of the parameter combination is adjusted and optimized to obtain the final combination of control parameters.

[0010] The final control parameter combination is applied to the servo motor system in real time, system feedback response data is collected, and stability indicators are analyzed. Based on the distribution of the stability indicators, the system adaptability is judged and an operating status evaluation result is generated.

[0011] Secondly, this application provides a servo motor adaptive parameter adjustment system, the system comprising:

[0012] The acquisition unit is used to acquire environmental data during the operation of the servo motor system through multiple sensors, fuse the environmental data into a comprehensive environmental vector, filter noise from the comprehensive environmental vector, and generate a smoothed environmental feature set.

[0013] The feature analysis unit is used to calculate the deviation vector between the current working condition and the preset benchmark working condition based on the environmental feature set, determine the significant features within the environmental change period based on the deviation vector, and if the significant features exceed the preset threshold, extract specific data from the environmental feature set and calculate the parameter change rate range based on the specific data.

[0014] The mapping unit is used to obtain the parameter adjustment sequence under multiple working conditions according to the parameter change rate range, map the parameter adjustment sequence with historical operating data, calculate the parameter mapping error between the new working condition parameter set and the actual working condition, and if the parameter mapping error exceeds the preset tolerance, iteratively update the adaptive adjustment coefficient to obtain the optimized adjustment coefficient set.

[0015] The simulation unit is used to integrate the constraints in the set of adjustment coefficients, verify the consistency of parameters and boundary compliance under the constraints through simulation technology, adjust the weight distribution of parameter combinations and optimize to obtain the final combination of control parameters.

[0016] An adaptive adjustment unit is used to apply the final control parameter combination to the servo motor system in real time, collect system feedback response data, analyze stability indicators, judge the system adaptability based on the distribution of the stability indicators, and generate operating status evaluation results.

[0017] A third aspect of this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the aforementioned servo motor adaptive parameter adjustment method.

[0018] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0019] First, temperature, humidity, and voltage data are collected from multiple sensors and fused into a comprehensive environmental vector. Noise is filtered through signal smoothing, Fourier transform, and wavelet transform, and impact-related abnormal fluctuation signals are extracted to accurately separate key components of external interference. The operating condition deviation vector is calculated, and impact direction angle and duration-related feature values ​​are extracted. Statistical analysis and time series analysis are combined to determine significant features, enabling accurate identification of environmental changes and impact interference. Then, based on significant features, impact energy dissipation and contact area data are extracted. Linear interpolation and boundary analysis generate parameter change rate ranges, defining reasonable boundaries for parameter adjustment and avoiding over-correction or under-correction. Parameter adjustment sequences are mapped using parameter similarity scores and weight distributions from historical data. Equivalent transformation is used to adapt to new operating conditions, and root mean square error is used to quantify mapping errors, ensuring the targeted and accurate nature of parameter adjustments. Next, an adaptive adjustment model is constructed based on the least squares method. The adjustment coefficients are iteratively optimized using the gradient descent algorithm. Convergence analysis and multi-dimensional verification form an optimized coefficient set, improving the model's adaptability to complex operating conditions. Finally, the consistency of parameters and boundary compliance were verified through finite element simulation based on the integrated constraints. The parameter weight distribution was optimized using the gradient descent algorithm to determine the optimal combination of control parameters, ensuring parameter reliability. After applying the parameter combination, feedback data was collected, and the comprehensive stability index was calculated through averaging filtering and statistical analysis to accurately assess the system's adaptability, forming a closed loop of "optimization-application-evaluation-iteration." This significantly improves the control accuracy, response stability, and environmental adaptability of the servo motor under complex disturbances. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of a servo motor adaptive parameter adjustment method according to an embodiment of this application;

[0022] Figure 2 This is a schematic diagram illustrating parameter mapping error and gradient descent optimization in an embodiment of this application;

[0023] Figure 3This is a schematic diagram of the finite element simulation verification comparison curves of the embodiments of this application;

[0024] Figure 4 This is a schematic diagram of the structure of a servo motor adaptive parameter adjustment system according to an embodiment of this application. Detailed Implementation

[0025] This application provides a servo motor adaptive parameter adjustment method, system, and storage medium. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0026] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of a servo motor adaptive parameter adjustment method in this application includes:

[0027] Step S1: Collect environmental data of the servo motor system during operation through multiple sensors, fuse the environmental data into a comprehensive environmental vector, filter noise from the comprehensive environmental vector, and generate a smoothed environmental feature set.

[0028] The generated smoothed environmental feature set includes:

[0029] Environmental data includes temperature, humidity, and voltage data. After normalization, these data are combined to form a three-dimensional comprehensive environmental vector. A moving average filter is applied to smooth the data in each dimension of the comprehensive environmental vector, generating preliminary smoothed environmental feature data. Fourier transform and wavelet transform are then performed on the preliminary smoothed environmental feature data to identify abnormal fluctuation signals. The time-domain features and high-frequency components of the abnormal fluctuation signals are extracted to form a smoothed environmental feature set containing the impact frequency distribution characteristics. Short-time Fourier transform is used to perform time-frequency analysis on the abnormal fluctuation signals, selecting key components and quantizing them into vector form. These key component vectors are then fused into the smoothed environmental feature set.

[0030] Specifically, temperature, humidity, and voltage sensors are deployed at key locations in the servo motor system. The temperature and humidity sensors can be installed on the motor housing to monitor the ambient temperature and humidity during servo operation in real time. The voltage sensor is deployed on the control circuit board to collect real-time power supply voltage data during servo operation. The temperature sensor collects temperature data ranging from -20 to 80 degrees Celsius, covering the common operating temperature range of servo motors, ensuring that the collected data reflects temperature changes under actual operating conditions. The humidity sensor collects humidity data ranging from 0 to 100%, comprehensively capturing the impact of different humidity environments on the servo motor. The voltage sensor collects voltage data ranging from 5 to 24 volts, matching the nominal power supply voltage range of the servo motor system, accurately reflecting power supply stability. The collected temperature, humidity, and voltage data constitute environmental data, which are core parameters reflecting the operating environment of the servo motor. Temperature changes affect the resistance and lubrication performance of the motor inside the servo motor, humidity fluctuations may cause electronic components to become damp and affect insulation, and unstable voltage directly affects the driving force of the servo motor actuator. These three factors together constitute the main environmental factors affecting the stability of the servo motor control parameters.

[0031] Normalization is a data preprocessing method that maps data with different dimensions and value ranges to a unified interval, eliminating the weight imbalance caused by differences in data dimensions. In this embodiment, temperature and humidity data are mapped to the interval [0, 1], while voltage data is scaled according to the nominal voltage value of the servo motor. For example, if the nominal voltage of the servo motor is 12 volts, the actual collected voltage data is scaled proportionally to the interval [0, 1]. After normalization, the temperature, humidity, and voltage data are combined to form a three-dimensional comprehensive environmental vector. This comprehensive environmental vector is a unified representation of multi-source environmental data, with each dimension corresponding to the normalization result of an environmental parameter. This comprehensive environmental vector can intuitively and comprehensively reflect the current overall environmental state of the servo motor, avoiding the one-sidedness caused by the analysis of a single environmental parameter.

[0032] A moving average filter is used to smooth the data of each dimension of the comprehensive environmental vector. The moving average filter is a linear filtering method based on the time domain. Its core principle is to reduce the interference of random noise by averaging the data from multiple consecutive sampling points. Random noise refers to irregular fluctuations in environmental data caused by factors such as sensor accuracy limitations and electromagnetic interference during data acquisition. This noise can mask the true trend of environmental changes, and direct use in subsequent analysis can lead to misjudgments in parameter adjustments. For example, the window size of the moving average filter is set to 10 consecutive sampling points. This means that the arithmetic mean of 10 consecutive sampling points for each dimension of the comprehensive environmental vector is calculated, and this average is used as the smoothed data for the current sampling point. This process is repeated for all sampling points to complete the smoothing of each dimension, generating preliminary smoothed environmental feature data. Therefore, the input data is the original sampling data of each dimension of the comprehensive environmental vector, and the output data is the preliminary smoothed environmental feature data for each dimension. Its function is to offset the influence of random noise through averaging, making the trend of environmental changes clearer.

[0033] The Fourier transform is a mathematical transformation method that converts a time-domain signal into a frequency-domain signal. Its principle is based on the orthogonality of trigonometric functions, decomposing any time-domain signal into a superposition of sine and cosine waves of different frequencies. By analyzing the amplitude and phase of each frequency component in the frequency domain, the characteristic frequency components in the signal can be identified. Therefore, the input data is the pre-smoothed environmental characteristic data, and the output data is the corresponding frequency-domain signal. When a servo motor is subjected to external impact, the environmental data will exhibit fluctuations at specific frequencies, which are abnormal fluctuation signals related to the impact frequency distribution. These abnormal fluctuation signals will manifest as amplitude peaks within a specific frequency range in the frequency domain. Therefore, the Fourier transform can detect abnormal peaks within a frequency range of 1 to 50 Hz. This frequency range corresponds to the frequency range of common impact events for servo motors, and the frequency and amplitude of the peak reflect the frequency characteristics and intensity of the impact.

[0034] Wavelet transform is a time-frequency localization analysis method. Its principle is to decompose a signal into multiple scales by scaling and translating wavelet basis functions. It can analyze the frequency characteristics of a signal like Fourier transform while preserving its temporal information, effectively separating high-frequency and low-frequency components. In this embodiment, the time-domain signal corresponding to the abnormal peak identified by Fourier transform is used as the input data for wavelet transform. Wavelet transform decomposes this signal, separating the high-frequency components, which are the local abnormal fluctuations caused by the impact event. Since signals corresponding to normal environmental changes are mostly low-frequency and stable, while the fluctuations caused by impacts are instantaneous and high-frequency, wavelet transform can highlight these local anomalies related to the impact frequency distribution, thereby extracting the time-domain features of the abnormal fluctuation signal, including waveform amplitude changes and duration, while preserving the feature data corresponding to the high-frequency components. These time-domain features and high-frequency components are integrated with the initially smoothed environmental feature data to form a smoothed environmental feature set containing the impact frequency distribution characteristics. This environmental feature set preserves the overall trend of environmental data changes while highlighting abnormal features related to external impacts, achieving in-depth mining of environmental data.

[0035] The Short-Time Fourier Transform (SFT) introduces a sliding time window into the Fourier Transform. Its principle is to divide a long-time-domain signal into multiple short-segment signals, perform a Fourier Transform on each segment, and thus obtain the frequency domain characteristics of the signal within different time windows, forming a spectrum on the time-frequency plane. This method overcomes the limitation of traditional Fourier Transform in simultaneously reflecting the time and frequency information of a signal, and can accurately locate the frequency changes of abnormal fluctuation signals within different time periods. In this embodiment, the input data for the SFT is the abnormal fluctuation signal. The size of the sliding time window is set to balance time resolution and frequency resolution, accurately capturing the instantaneous frequency changes of the signal while ensuring the accuracy of frequency analysis. The output data is the time-spectrum graph of the abnormal fluctuation signal.

[0036] Critical components refer to frequency components related to external interference such as shocks and electromagnetic interference. They are determined using an energy threshold screening method. For example, the energy threshold is set to 1.5 times the average energy of all frequency components of the signal. Frequency components with energy exceeding this threshold are identified as critical components because external interference causes the energy of the corresponding frequency components to be significantly higher than that produced by normal environmental changes. The selected critical components are then quantized into vector form. The quantization process converts the characteristic parameters of the critical components, such as amplitude, frequency, and duration, into numerical vectors. This vector is the critical component vector. Finally, this critical component vector is fused into a smoothed environmental feature set. The fusion method is vector concatenation, where the critical component vector is added as a new dimension to the original environmental feature set. This ensures that the final environmental feature set not only includes basic environmental information such as temperature, humidity, and voltage, but also integrates key features related to external interference. This provides comprehensive and accurate basic data input for subsequent calculations of the deviation between the current operating condition and the baseline operating condition, and for analyzing the significant characteristics of environmental changes. This ensures that subsequent parameter adjustments can specifically address the impact of external shocks and environmental changes.

[0037] Step S2: Calculate the deviation vector between the current working condition and the preset benchmark working condition based on the environmental feature set. Determine the significant features within the environmental change period based on the deviation vector. If the significant features exceed the preset threshold, extract specific data from the environmental feature set and calculate the parameter change rate range based on the specific data.

[0038] Among them, identifying the salient features within the period of environmental change includes:

[0039] Based on the standard parameter vector of the preset benchmark working condition, an element-wise subtraction operation is performed with the current working condition vector corresponding to the smoothed environmental feature set to obtain the deviation vector; the components corresponding to the impact direction angle and impact duration are separated from the deviation vector, and the main frequency and amplitude peak of the components are extracted as feature values ​​through Fourier transform; the significant features within the environmental change period are determined based on the standard deviation of the feature values.

[0040] The range of the rate of change of the calculated parameters includes:

[0041] From the smoothed environmental feature set, numerical sequences reflecting impact energy attenuation and impact contact area data containing contact surface size and distribution density are selected as specific data. Outliers in the specific data are filtered and arranged in time series to form a discrete point set. The discrete point set is then interpolated and smoothed sequentially to obtain a parameter change rate sequence. The maximum value in the parameter change rate sequence is identified as the upper limit of the parameter change rate, and the parameter change rate range is formed based on the upper limit of the parameter change rate.

[0042] Specifically, the preset baseline operating condition standard parameter vector is a predefined set of standard values ​​based on the historical stable operating data of the servo motor. This set includes temperature, humidity, and voltage baseline values. This standard parameter vector serves as a reference for assessing whether the current operating condition is normal, ensuring that parameter adjustments have a clear basis for comparison. The current operating condition vector corresponding to the smoothed environmental feature set refers to the vector data containing the current operating status of temperature, humidity, and voltage, formed after data acquisition, fusion, and noise filtering in step S1. Each dimension of this vector corresponds one-to-one with the dimensions of the standard parameter vector.

[0043] In this embodiment, the element-by-element subtraction operation refers to subtracting the data of each dimension in the current operating condition vector from the benchmark value of the corresponding dimension in the standard parameter vector. The difference obtained forms a deviation vector, which intuitively reflects the degree of deviation of the current operating condition from the benchmark operating condition. The value of each element represents the deviation of the corresponding environmental parameter, and the sign indicates the direction of deviation. Positive values ​​are higher than the benchmark, and negative values ​​are lower than the benchmark, clearly showing the deviation of each environmental parameter.

[0044] First, the components corresponding to the impact direction angle and impact duration are separated from the deviation vector. The impact direction angle refers to the angular deviation between the actual impact direction and the expected design direction when the servo motor is subjected to an external impact, usually measured in degrees. This component is separated from the deviation vector using a vector projection method, which projects the deviation vector onto a preset impact direction coordinate system to extract the component data related to the impact direction. The impact duration refers to the time span from the occurrence to the end of the external impact, measured in milliseconds. Its corresponding component is obtained by extracting data from the deviation vector using a time window function. The time window function sets a fixed time interval and extracts the data segment related to the impact duration from the deviation vector within that time interval as the corresponding component.

[0045] Subsequently, the dominant frequency and amplitude peak of the two components are extracted as feature values ​​through Fourier transform. Therefore, the input data of the Fourier transform are the time-domain signals corresponding to the separated impact direction angle component and impact duration component, and the output data are the frequency-domain signals corresponding to these two components. The dominant frequency of the impact direction angle component and the amplitude peak of the impact duration component can be directly read from the frequency-domain signal. The dominant frequency refers to the frequency component with the largest amplitude in the frequency-domain signal, which reflects the periodicity of the impact. The amplitude peak is the largest amplitude value in the frequency-domain signal, which corresponds to the intensity of the impact. These two parameters together constitute the feature values ​​that can characterize the impact characteristics.

[0046] The statistical mean and standard deviation of the extracted feature values ​​(dominant frequency and peak amplitude) are calculated. The mean is the arithmetic mean of all feature values, reflecting the central tendency. The standard deviation is the square root of the average of the squares of the deviations of each feature value from the mean, reflecting the dispersion of the feature values. In this embodiment, significant features refer to outliers corresponding to feature values ​​exceeding twice the standard deviation. The principle behind this criterion is that in a normally distributed dataset, approximately 95% of the data will fall within the range of the mean plus or minus two standard deviations. Data points exceeding this range are low-probability anomalies and can be identified as features that significantly deviate from the normal state, i.e., significant features related to external shocks. Through this screening process, abnormal features caused by external shocks can be accurately identified from the feature values, eliminating interference from normal environmental fluctuations.

[0047] Specifically, the numerical sequence of impact energy attenuation refers to the continuous data sequence formed during the transmission and dissipation of impact energy within the servo system after the servo motor is subjected to an external impact. This sequence directly reflects the changing trend of impact energy. The impact contact area data refers to the geometric characteristic data of the contact part between the external impact and the servo motor. The contact surface dimensions include physical parameters such as the length and width of the contact area, while the distribution density is the degree of distribution of impact stress within the contact area. These specific data can reflect the physical range and intensity distribution of the impact on the servo motor. Together, they serve as the core data for calculating the range of parameter change rate, ensuring that the parameter change rate calculation closely matches the actual impact of the impact on the servo motor.

[0048] Outliers are isolated data points that deviate from the normal data trend due to factors such as sensor errors and data transmission interference. The filtering method uses a preset threshold method, which sets a normal range for impact energy attenuation and impact contact area. Data points exceeding this range are identified as outliers and removed to ensure that the retained specific data accurately reflects the actual impact-related situation. The filtered specific data is then arranged into a discrete point set according to a time series. Time series arrangement means sorting the data according to the chronological order of data acquisition, so that the data can show the change pattern over time. The discrete point set refers to the data showing a discontinuous distribution on the time axis, requiring further processing to obtain a continuous trend.

[0049] Interpolation and smoothing are performed sequentially on the discrete point set. Linear interpolation is used for interpolation, which is a method to calculate missing intermediate data by connecting adjacent data points with straight lines. The principle is that the change between two adjacent data points is assumed to be linear. For example, for two adjacent data points A(t1, m1) and B(t2, m2), where t is time and m is the data value, the intermediate data value m corresponding to any time t (t1 < t < t2) satisfies m = m1 + (m2 - m1) * (t - t1) / (t2 - t1). This method can fill in the blank data between discrete point sets, forming a continuous parameter sequence, and solving the problem that discrete data cannot reflect continuous changing trends. The continuous parameter sequence obtained after interpolation may still contain a small amount of noise interference, so smoothing is required. A moving average filter can be used for smoothing. The principle of the moving average filter is to calculate the arithmetic mean of multiple data points within a fixed window size for the continuous parameter sequence, and use this mean as the smoothed value of the center data point of the window. The window is traversed sequentially to complete the smoothing process. By setting the window size, this process can further reduce noise interference and obtain a stable parameter change rate sequence.

[0050] The parameter change rate refers to the magnitude of parameter change per unit time, reflecting the speed of parameter change. Its calculation formula is the ratio of the parameter change to the time change. By traversing all data points in the parameter change rate sequence, the data point with the largest value is selected as the upper limit of the parameter change rate. This upper limit represents the maximum possible speed of parameter change under the current impact condition. Based on this upper limit, a parameter change rate range is formed. Specifically, the upper bound of the parameter change rate range is set to the aforementioned upper limit, and the lower bound is 0.5 times the upper limit, forming a closed interval parameter change rate range. This parameter change rate range sets reasonable boundary constraints for subsequent servo motor parameter adjustments, avoiding excessive parameter adjustments that could lead to system oscillation or insufficient adjustments that could lead to system instability, ensuring that parameter adjustments are performed within a safe and effective range.

[0051] Step S3: Obtain the parameter adjustment sequence under various operating conditions based on the parameter change rate range, map the parameter adjustment sequence with historical operating data, calculate the parameter mapping error between the new operating condition parameter set and the actual operating condition, and if the parameter mapping error exceeds the preset tolerance, iteratively update the adaptive adjustment coefficient to obtain the optimized adjustment coefficient set.

[0052] The calculation of parameter mapping error includes:

[0053] Based on historical operating data, a parameter similarity score is calculated between historical operating condition parameters and the current parameter change rate range. The influence ratio of different parameters under variable operating conditions is extracted as the parameter weight distribution. A mapping relationship is constructed based on the parameter similarity score and parameter weight distribution. Based on the mapping relationship, the historical parameter sequence that best matches the current parameter change rate range is selected to form a parameter adjustment sequence under variable operating conditions. The impact components corresponding to the impact direction angle and impact duration in the parameter adjustment sequence are identified. Based on the similarity transformation matrix of the deviation vector, the impact components are linearly mapped to the new operating condition coordinate system, generating a new operating condition parameter set that includes the impact force peak range and damping coefficient. The root mean square error formula is used to compare the differences between specific data in the new operating condition parameter set and the actual collected feedback response data to generate the parameter mapping error.

[0054] The optimized set of adjustment coefficients includes:

[0055] When the parameter mapping error exceeds the preset tolerance range, an adaptive adjustment model is constructed based on the new operating condition parameter set; the adjustment coefficients in the adaptive adjustment model are iteratively adjusted using the gradient descent algorithm until the error is minimized; the variance of the adjustment coefficients after iteration is calculated, and if the variance is less than the first preset value, the coefficients are determined to be converged, and the converged adjustment coefficients are summarized to form an initial adjustment coefficient set; test data is extracted to verify the applicability of the initial adjustment data coefficient set, and coefficient combinations are selected based on the applicability as the optimized adjustment coefficient set.

[0056] The construction of the adaptive adjustment model includes:

[0057] Effective temperature data and dynamic adjustment rate data were selected from the new operating condition parameter set. After removing outliers, the data were normalized. A feature sample set was constructed using the normalized effective temperature data as the independent variable and the normalized dynamic adjustment rate data as the dependent variable. A linear model expression between the effective temperature range and the dynamic adjustment rate was defined. An error sum of squares function was defined based on the least squares method. By taking the partial derivative of the error sum of squares function, a system of equations was obtained. The statistics of the feature sample set were calculated and substituted into the system of equations to obtain the optimal solution of the coefficients of the linear model expression. An initial adaptive adjustment model was formed based on the optimal solution of the coefficients. The initial adaptive adjustment model was verified using leave-one-out cross-validation. The mean absolute error and root mean square error of all test samples were calculated. If the mean absolute error and root mean square error both met the corresponding preset error range, the fit was deemed valid; otherwise, the data processing method was adjusted and refitting was performed. The model coefficients of the initial adaptive adjustment model were converted inversely to the corresponding coefficients of the actual physical quantities. The upper and lower limits of the dynamic adjustment rate were set in combination with the servo motor operation constraints to form an adaptive adjustment model.

[0058] Specifically, historical operating data is a complete data set recorded by the servo motor under different operating conditions in the past. It includes environmental parameters, control parameters, and operating status data corresponding to each operating condition, and serves as the basis for drawing on past effective adjustment experience. Parameter similarity score is an indicator used to quantify the degree of matching between historical operating conditions and current operating conditions. Its calculation is implemented through the Euclidean distance algorithm. Euclidean distance refers to the straight-line distance between two points in multidimensional space. The historical operating condition parameters and the current parameter change rate range are regarded as two points in multidimensional space, respectively. By calculating the Euclidean distance between them, the smaller the distance, the higher the parameter similarity score, and vice versa. Therefore, the input data are the historical operating condition parameter vector and the current parameter change rate range vector, and the output data is the corresponding Euclidean distance value. Then, through a preset mapping rule, the distance value is converted into a parameter similarity score in the range of 0-1. The closer the score is to 1, the higher the matching degree between the historical operating conditions and the current operating conditions.

[0059] Parameter weight distribution refers to the proportion of influence of different control parameters on the operational stability of a servo motor under varying operating conditions. For example, the weight of parameters related to impact energy dissipation might be set to 0.4, the weight of parameters related to impact contact area to 0.3, and the weight of parameters related to voltage stability to 0.3. These weights are determined based on extensive experimental data and engineering experience, reflecting the importance of different parameters in actual operation. A mapping relationship is constructed by multiplying the similarity score by the weight of each parameter using a weighted average algorithm and summing the results to obtain a comprehensive matching index. Then, based on this comprehensive matching index, historical parameter sequences that best match the current parameter change rate range are selected from historical operating data. These historical parameter sequences represent parameter adjustment schemes that have been proven effective in practice under similar past operating conditions. Using these sequences as parameter adjustment sequences under varying operating conditions ensures that subsequent parameter adjustments have a reliable historical basis.

[0060] The impact direction deviation angle refers to the angle between the actual impact direction and the expected design direction when the servo motor is subjected to an external impact, measured in degrees. The impact duration refers to the time span from the occurrence to the end of the impact, measured in milliseconds. These two parameters are core indicators characterizing impact characteristics, and their corresponding impact components directly affect the parameter adjustment strategy of the servo motor. The similarity transformation matrix of the deviation vector is a linear transformation matrix constructed based on the similarity of the deviation vectors between historical and new operating conditions. The deviation vector is the parameter difference vector between the current operating condition and the reference operating condition. The similarity calculation uses the cosine similarity algorithm. The cosine similarity measures the degree of similarity between two vectors by calculating the cosine value of the angle between them, with a value ranging from -1 to 1. The closer to 1, the higher the similarity. The elements of this transformation matrix are the cosine similarity values ​​under different parameter dimensions. Its function is to linearly map the impact components in the historical parameter adjustment sequence to the new operating condition coordinate system, realizing the adaptation and transformation between historical parameters and the new operating condition. Linear mapping refers to converting the vector of impact components into an equivalent parameter vector under the new operating condition through matrix multiplication. This ensures that the mapped parameters retain the effective characteristics of the historical adjustment sequence while adapting to the operational requirements of the new operating condition. Ultimately, it generates a set of parameters for the new operating condition, including the peak impact force range and the damping coefficient. The peak impact force range is the maximum impact force range that the servo motor can withstand, and the damping coefficient is a parameter that characterizes the damping characteristics of the servo motor system. Together, they constitute the core reference for parameter adjustment under the new operating condition.

[0061] The root mean square error (RMSE) is a commonly used metric for quantifying data deviation. Its principle involves first calculating the difference between specific data in the new operating condition parameter set for each parameter dimension and the actual collected feedback response data. The difference is then squared, and the average value is calculated. Finally, the square root of the average value is taken to obtain the final error value. This effectively reflects the overall degree of deviation between the two sets of data. The input data consists of the impact peak force, damping coefficient, and the corresponding actual collected feedback response data corresponding to specific parameter data in the new operating condition parameter set. The output data is a single RMSE value, which is the parameter mapping error, used to determine the degree of fit between the new operating condition parameter set and the actual operating condition.

[0062] For example, Figure 2 For the parameter mapping error versus gradient descent optimization graph, in Figure 2The upper subplot illustrates the convergence process of the parameter mapping error with the number of iterations. The curves in the figure show that as the gradient descent algorithm iterates, the parameter mapping error decays exponentially, gradually decreasing to below the preset error threshold line (dashed line in the figure), demonstrating the effectiveness of the gradient descent algorithm in minimizing error. The lower subplot shows the convergence process of the three key adjustment coefficients with the number of iterations. Curves of different gray levels and line types represent the trajectories of adjustment coefficient 1, adjustment coefficient 2, and adjustment coefficient 3, respectively. This process intuitively verifies that the gradient descent algorithm can effectively promote the coordinated convergence of multi-dimensional adjustment coefficients and avoid system instability caused by coefficient oscillations.

[0063] Specifically, effective temperature data refers to temperature parameters within the normal operating temperature range of the servo motor. Dynamic adjustment rate data refers to the magnitude of parameter change per unit time, for example, a parameter change rate close to 0.5 units per second. These data are key characteristics affecting the adaptability of servo motor parameter adjustment. Outlier removal is performed on the selected data. Outliers are isolated points that deviate from the overall trend of the data. The 3σ criterion is used for identification, that is, when the data value exceeds the mean plus or minus 3 times the standard deviation, it is judged as an outlier and removed to ensure the reliability of the data. The processed effective temperature data and dynamic adjustment rate data are normalized respectively. Normalization is a preprocessing method that maps the data to the [0, 1] interval. Its purpose is to eliminate the influence of differences in the dimensions of different data, so that the two types of data have the same weight priority in the subsequent model fitting. The input data for this process are the original effective temperature data and dynamic adjustment rate data, and the output data is the normalized standardized data. Using normalized effective temperature data as the independent variable x and normalized dynamic adjustment rate data as the dependent variable y, a feature sample set is constructed, which forms several pairs of sample data (x_i, y_i), where i is the sample number, to provide basic data for fitting the adaptive adjustment model.

[0064] The linear model expression is defined as: y = ax + b. This expression assumes a linear correlation between the effective temperature range and the dynamic adjustment rate, where a is the slope coefficient and b is the intercept coefficient, both of which are model parameters to be solved. Based on the least squares method, the error sum of squares function is defined as: S = Σ(y_i - (ax_i + b))² (i from 1 to n, where n is the number of samples). The least squares method is a mathematical method for finding the optimal solution for model parameters. Its core principle is to adjust the values ​​of parameters a and b to minimize the error sum of squares function S, at which point the corresponding parameters are the optimal solution. The essence of this method is to minimize the overall deviation between the model's predicted values ​​and the actual sample values. Substituting the optimal solution into the linear model expression completes the fitting of the initial adaptive adjustment model. The input data of this initial adaptive adjustment model is the normalized effective temperature data, and the output data is the normalized predicted dynamic adjustment rate value. Its working principle is to correlate temperature changes with the parameter dynamic adjustment rate through a linear relationship, achieving adjustment rate prediction based on temperature conditions.

[0065] Leave-one-out cross-validation is a model validation method. Its principle is to use each sample in the feature set as the test set and the remaining samples as the training set to refit the model. The refitted model is then used to predict the test samples, and the deviation between the predicted and actual values ​​is calculated. This method can fully utilize limited sample data to comprehensively evaluate the model's generalization ability. Its input data is the complete feature set, and the output data is the prediction deviation for each test sample. Based on the prediction deviations of all test samples, the mean absolute error (MAE) and root mean square error (RMSE) are calculated. The mean absolute error is the average of the absolute values ​​of all prediction deviations, used to measure the average deviation between the predicted and actual values. The root mean square error is the square root of the average of the sum of the squares of all prediction deviations, and is more sensitive to larger deviations, highlighting the impact of extreme errors. Both are core indicators for measuring the model's prediction accuracy. For example, the threshold for the mean absolute error is preset to 0.05 and the threshold for the root mean square error is 0.08. If the calculated mean absolute error is ≤0.05 and the root mean square error is ≤0.08, the model is considered to be fit effectively. If the condition is not met, the process returns to the data preprocessing stage, the normalization method is adjusted or the screening range of the dynamic adjustment rate data is expanded, and the feature sample set is reconstructed and the model is fitted again until the error requirements are met.

[0066] After validating the model, the coefficients 'a' and 'b' of the initial adaptive adjustment model are inversely converted to the coefficients corresponding to the actual physical quantities. This inverse conversion involves performing the inverse operation of the normalization process to restore the coefficients obtained from standardized data to their original counterparts. This allows the initial adaptive adjustment model to directly receive the original temperature data and output predicted values ​​of the dynamic adjustment rate in a practically meaningful way. Simultaneously, considering the operational constraints of the servo motor, upper and lower limits for the dynamic adjustment rate are set. These constraints are determined based on actual engineering parameters such as the mechanical structural strength and the withstand capability of electronic components, ensuring that the dynamic adjustment rate output by the initial adaptive adjustment model remains within the safe operating range of the servo motor. This ultimately forms an adaptive adjustment model that can be used for subsequent iterative calculations.

[0067] Specifically, gradient descent is a commonly used optimization algorithm. It calculates the gradient direction of the loss function and updates the model parameters progressively in the opposite direction of the gradient, thereby minimizing the loss function value. The loss function is defined as the sum of squared deviations between the predicted parameter adjustment effect and the actual desired effect. The input data for this algorithm includes the initial adjustment coefficients of the adaptive adjustment model, the loss function, and a preset learning rate. The learning rate is a key parameter controlling the step size of parameter updates, used to balance iteration speed and convergence accuracy. The output data is the adjustment coefficients updated after each iteration. During the iteration process, after each update of the adjustment coefficients, the corresponding loss function value is calculated. It is then determined whether the loss function value has reached a preset minimum threshold or whether the number of iterations has reached a preset upper limit. If either condition is met, the iteration stops, and the adjustment coefficient at this point is the current optimal value. This process dynamically optimizes the adjustment coefficients, improving the adaptation accuracy of the adaptive adjustment model to new parameter sets. Variance is a statistic that measures the dispersion of data. It is used to determine whether the adjustment coefficients have converged. If the variance is less than a first preset value, it indicates that the adjustment coefficients have stabilized during the iteration process without significant fluctuations, and the coefficients are considered to have converged. If the variance is greater than or equal to the first preset value, the learning rate or number of iterations of the gradient descent algorithm needs to be readjusted, and the iteration calculation needs to be performed again until the coefficients converge. All converged adjustment coefficients are summarized to form an initial set of adjustment coefficients. This initial set of adjustment coefficients represents the preliminary optimization results obtained based on the current model and data.

[0068] Test data is selected from a database of typical operating conditions for servo motors, covering parameter data under various extreme and conventional conditions, including low temperature, high temperature, high humidity, and voltage fluctuations. Suitability refers to the degree to which the initial set of adjustment coefficients conforms to the preset stable operating standard when applied to different test conditions. This is calculated by applying the initial set of adjustment coefficients to each test condition, simulating the servo motor's operating state, calculating the impact response delay and parameter fluctuation amplitude corresponding to the adjusted system stability indicators, and comparing them with the preset pass standard. If a certain coefficient combination can make the system stability indicators meet the pass standard in more than 80% of the test conditions, then the suitability of that coefficient combination is deemed satisfactory. Based on the suitability evaluation results, the coefficient combination that meets the standard and has the best overall performance is selected as the optimized adjustment coefficient set. This set of adjustment coefficients can adapt to various complex operating conditions, providing a precise and reliable core basis for servo motor parameter adjustment, ensuring that the servo motor maintains a stable operating state under changing conditions.

[0069] Step S4: Integrate the constraints of the adjustment coefficient set, verify the parameter consistency and boundary compliance under the constraints through simulation technology, adjust the weight distribution of the parameter combination and optimize to obtain the final control parameter combination.

[0070] The optimized final control parameter combination includes:

[0071] The impact force peak range and impact damping coefficient interval are selected from the set of adjustment coefficients as constraints. The average value of the impact force peak is calculated as the verification benchmark, and the constraints are transformed into vector form. A finite element simulation model is constructed, and the vectors corresponding to the constraints are input into the finite element simulation model and a virtual load is applied. Multiple iterative simulations are run, and the parameter consistency index is calculated by comparing the percentage deviation between the simulated output and the expected output to check whether the impact damping coefficient is within the preset range. Based on the parameter consistency index and boundary violation rate in the simulation verification report, a weight adjustment factor is obtained through linear mapping, and the weight distribution of the parameter combination is updated based on the weight adjustment factor. The gradient descent algorithm is used to optimize the weight distribution using the sum of squares of the deviations between the weight distribution and the verification results as the loss function. The parameter value corresponding to the minimum loss is selected based on the loss function to form the final control parameter combination.

[0072] Specifically, the adjustment coefficient set is a set of coefficients adapted to different operating conditions obtained after iterative optimization. It includes the peak impact force range and impact damping coefficient range, which are directly related to the servo motor's impact resistance performance; these are the constraints. The peak impact force range refers to the maximum impact force range that the servo motor can withstand during normal operation. Its value needs to be determined in conjunction with engineering parameters such as the servo motor's mechanical structural strength and material tolerance limits. This range ensures that the servo motor can cope with normal impact interference while avoiding damage to internal components due to excessive impact force. The impact damping coefficient range is a parameter range characterizing the damping characteristics of the servo motor system. The damping coefficient is used to buffer impact energy and suppress system oscillations. This range allows the servo motor to quickly recover to a stable state after an impact, avoiding response delays or jitter. The average peak impact force is calculated using the arithmetic mean method: (lower limit of peak impact force + upper limit of peak impact force) / 2. This verification benchmark is used in subsequent simulation verification to determine whether the simulated output is within a reasonable range. To facilitate the input and calculation of the subsequent finite element simulation model, the extracted constraints are converted into vector form. Specifically, the upper and lower limits of the impact force peak range, the upper and lower limits of the impact damping coefficient range, and the average value of the impact force peak are arranged in a preset order to form a four-dimensional constraint vector. This vector is a mathematical representation of the constraints and can be directly identified and processed by the simulation model to ensure the accurate transmission of the constraints.

[0073] The finite element simulation model is a mathematical model built based on the finite element analysis method to simulate the operating state of a servo motor. The finite element analysis method decomposes the complex servo motor system structure into several interconnected micro-units (i.e., finite elements). By solving the stress and strain equations of each unit, the mechanical response characteristics of the entire system are obtained by superimposing them. This method can accurately simulate the operating state of the servo motor under different load and constraint conditions. Its input data are the structural parameters, material properties, and transformed constraint vectors of the servo motor, and the output data is the simulated operating response results of the system.

[0074] The virtual load simulates the external impact load that the servo motor system may encounter during actual operation. Its magnitude and direction are set based on common impact scenarios. To ensure the reliability of the verification results, multiple iterative simulations are run, with the number of iterations set to 30. During each iteration, the finite element simulation model calculates the system response under different parameter combinations based on the constraints.

[0075] For example, Figure 3The finite element simulation verification comparison curves show the relationship between simulation time and output value. The expected output curve (dark gray solid line) is based on the design specifications of the servo motor and historical stable operation data, representing the performance standard that the system should achieve under ideal conditions. The simulation output curve (gray dashed line) is the actual simulation result calculated by the finite element simulation model after applying virtual load and input constraint vector. The error region (light gray fill) intuitively shows the deviation range between the simulation output and the expected output. Positive error (simulation output higher than expected) and negative error (simulation output lower than expected) are represented by different gray levels, which verifies the effectiveness of the finite element simulation model in the parameter adjustment verification stage.

[0076] The parameter consistency index is calculated by comparing the percentage deviation between the simulated output and the expected output. The expected output is the ideal response result preset based on the servo motor design specifications and historical stable operation data. The formula for calculating the percentage deviation is: (|simulated output value - expected output value| / expected output value) × 100%. The parameter consistency index is 1 minus this percentage deviation. The closer the index is to 1, the higher the fit between the simulated output and the expected output, and the better the parameter consistency. For example, when the percentage deviation is 5%, the corresponding parameter consistency index is 0.95, indicating that the parameters have high consistency under this condition. Simultaneously, during each iteration of the simulation, the impact damping coefficient is checked in real time to see if it is within the preset range. If it exceeds this range, it is marked as a boundary violation. The ratio of the number of boundary violations in all iterations to the total number of iterations is calculated to obtain the boundary violation rate. The boundary violation rate is used to measure whether the parameters meet the preset boundary constraint requirements. The simulation verification report includes the parameter consistency index and boundary violation rate for each iteration, providing data support for subsequent weight adjustments.

[0077] The weight adjustment factor is a coefficient used to adjust weights, calculated based on the parameter consistency index in the simulation verification report. Its calculation employs a linear mapping method, which maps the parameter consistency index from its original range to a preset weight adjustment factor range. This mapping relationship is determined based on engineering experience and extensive experimental data. For example, when the parameter consistency index is 0.95, the corresponding weight adjustment factor is 1.2; when the parameter consistency index is 0.9, the corresponding weight adjustment factor is 1.1. That is, the higher the parameter consistency, the larger the weight adjustment factor, and the greater the corresponding parameter weight will be. The weight distribution of the parameter combination refers to the proportion of the impact peak force and impact damping coefficient corresponding to different control parameters in the final control parameter combination. A larger weight indicates a more significant impact of that parameter on the stability of the servo motor. The weight distribution update process involves multiplying the initial weight of each parameter by the corresponding weight adjustment factor, and then normalizing the updated weights to ensure that the sum of the weights of all parameters is 1. Normalization is performed to avoid imbalance in the total weight due to weight adjustments and to ensure a reasonable proportion of each parameter's weight.

[0078] The gradient descent algorithm optimizes the weight distribution by calculating the gradient direction of the loss function and progressively updating the parameter weights along the inverse direction of the gradient. The loss function is defined as the sum of squares of the deviations between the weight distribution and the simulation results. The deviation refers to the difference between the simulated response and the expected response corresponding to the parameter combination under the current weight distribution. Calculating the sum of squares amplifies the impact of larger deviations, allowing the optimization process to focus more on improving key deviation issues. Therefore, the input data for the gradient descent algorithm consists of the updated weight distribution, the loss function, and a preset learning rate. The learning rate is a key parameter controlling the step size of the weight updates, and the output data is the weight distribution updated after each iteration. An upper limit of 50 iterations is set. In each iteration, the loss function value and gradient direction corresponding to the current weight distribution are calculated, and the weights of each parameter are adjusted according to the gradient direction until the loss function value reaches a preset minimum threshold or the maximum number of iterations is reached. The resulting weight distribution is the optimal weight distribution.

[0079] Within the range of peak impact force and impact damping coefficient, specific parameter values ​​that match the optimal weight distribution and minimize the loss function are selected. These parameter values ​​together constitute the final control parameter combination. The final control parameter combination is the optimal solution after comprehensively considering constraints, simulation verification results, and weight optimization. It can ensure that the servo motor maintains a stable operating state and high control accuracy when facing external impacts and environmental changes.

[0080] Step S5: Apply the final control parameter combination to the servo motor system in real time, collect system feedback response data, analyze stability indicators, judge system adaptability based on stability indicator distribution, and generate operating status evaluation results.

[0081] The generated operational status assessment results include:

[0082] An average filtering method is used to filter noise from the system feedback response data, extracting the millisecond-level impact response delay from the occurrence of the impact to the system stabilization, and calculating the historical comparison value of the parameter between the current impact response delay and the historical average delay. The average value of the impact response delay is calculated, and the variance of the parameter historical comparison value is calculated. The average value and variance are calculated using a weighted summation formula to obtain the comprehensive stability index. Based on the distribution of the comprehensive stability index, the adaptability of the servo motor system is judged, and the operating status evaluation result is generated.

[0083] Specifically, the final control parameter combination is the optimal set of parameters obtained through constraint fusion, simulation verification, and weight optimization. This set includes specific values ​​such as peak impact force and impact damping coefficient, which are directly related to the servo motor's shock resistance and operational stability. During application, these parameter values ​​are directly mapped to the servo motor's control module. Upon receiving the parameters, the control module immediately updates the system's operational configuration and simultaneously monitors the servo motor's operating status after parameter application. This ensures that the parameter switching process is conflict-free and anomalies-free, avoiding issues such as servo motor jitter and response delays caused by sudden parameter changes, thus achieving a smooth transition from optimized parameters to actual operation.

[0084] System feedback response data is real-time operational status data collected by temperature, humidity, and voltage sensors deployed in key locations such as the motor housing and control circuit board when the servo motor is running under new parameter configurations. This data accurately reflects the actual performance of the system after parameter adjustments, including key information such as response speed after an impact and operational stability. Due to potential electromagnetic interference and sensor accuracy limitations during data acquisition, random noise may be present in the data. This noise can interfere with the accuracy of subsequent feature extraction and index calculation. Therefore, an averaging filter method is needed to filter noise from the system feedback response data. The averaging filter method is a basic and effective time-domain smoothing filtering technique. Its core principle is to calculate the arithmetic mean of data from multiple consecutive sampling points to offset the influence of random noise and highlight the true trend of data changes. For example, setting the average filter window size to 5 consecutive sampling points means summing 5 consecutive sample values ​​for each dimension of the feedback response data and dividing by 5 to obtain the smoothed data at the center sampling point of the window. The window is then traversed sequentially across the entire feedback response dataset to complete the noise filtering process and output the smoothed response feature set. Therefore, the input data is the original system feedback response data, and the output data is the smoothed response feature set. Noise suppression is achieved through local averaging of the data.

[0085] Impact response delay refers to the time required for a servo motor system to return to a stable operating state after an external impact occurs, usually measured in milliseconds. It reflects the system's rapid response capability to impact interference; the shorter the delay, the better the system's impact resistance. During the extraction process, the difference between the starting time of the impact and the time when the system operating parameters tend to stabilize is calculated by identifying the response feature set. This difference is the impact response delay. Subsequently, a historical parameter comparison value is calculated. This historical comparison value is the difference between the currently extracted impact response delay and the historical average delay. The historical average delay is the arithmetic mean of multiple impact response delays during stable operation of the servo motor under similar past operating conditions. This comparison value allows for a direct assessment of the system's response speed under the current parameter configuration compared to historical levels. A positive value indicates that the current delay is higher than the historical average, and the system response is slower; a negative value indicates that the current delay is lower than the historical average, and the system response is faster. The comprehensive stability index is the core evaluation parameter that integrates the impact response delay and the historical parameter comparison value, comprehensively reflecting the system's operational stability. Specifically, this includes: First, calculating the average value of the extracted multiple impact response delay data points. The average value, or arithmetic mean, is the sum of all impact response delay data points divided by the number of data points, reflecting the overall level of the system's impact response delay and avoiding the influence of randomness from a single data point. Then, calculating the variance of the historical comparison values ​​of multiple parameters. Variance is a statistic that measures the degree of data dispersion. It is calculated by summing the squared differences between each historical comparison value and the average of those comparison values, then dividing by the number of data points minus 1. A smaller variance indicates less fluctuation in the historical comparison values ​​of the parameter and better consistency in the system's response speed. Finally, a weighted summation formula is used to calculate the above average value and variance to obtain the comprehensive stability index. The weighted summation formula is: Comprehensive Stability Index = × Average shock response delay + ×variance of historical comparison values ​​of parameter and All are weight values. The allocation of weight values ​​is determined based on the degree of influence of the two indicators on system stability. The two indicators are organically integrated through weighted summation, so that the comprehensive stability index can fully and balancedly reflect the operating status of the system.

[0086] The adaptability of a servo motor system refers to the degree to which the current final control parameter set is suitable for the servo motor's operating conditions. In the judgment process, a reasonable distribution range of comprehensive stability indicators is first preset. This reasonable distribution range is determined based on the servo motor's design specifications, historical stable operating data, and engineering experience. The percentage of all calculated comprehensive stability indicators falling within the reasonable range is then calculated. If this percentage reaches or exceeds the preset proportion, the system is considered to have high adaptability, indicating that the current parameter set is well-suited to the operating conditions and the system is operating stably. Otherwise, the system is considered to have average or low adaptability, suggesting that there is still room for optimization in the current parameter set.

[0087] The operational status assessment report not only includes the system adaptability level (high, medium, low), but also lists the specific calculation data of the comprehensive stability index, the average value of the impact response delay, the variance of the historical comparison values ​​of parameters, and other key information. It also puts forward targeted optimization suggestions for cases with insufficient adaptability, such as adjusting the impact damping coefficient and optimizing the peak impact force range. This provides clear and explicit data support for the further iterative optimization of the servo motor's adaptive parameters, forming a complete closed loop of "parameter optimization - application verification - evaluation feedback - re-optimization".

[0088] The above describes a servo motor adaptive parameter adjustment method according to an embodiment of this application. The following describes a servo motor adaptive parameter adjustment system according to an embodiment of this application. Please refer to [link to relevant documentation]. Figure 4 One embodiment of a servo motor adaptive parameter adjustment system in this application includes:

[0089] The acquisition unit is used to collect environmental data during the operation of the servo motor system through multiple sensors, fuse the environmental data into a comprehensive environmental vector, filter noise from the comprehensive environmental vector, and generate a smoothed environmental feature set.

[0090] The feature analysis unit is used to calculate the deviation vector between the current operating condition and the preset benchmark operating condition based on the environmental feature set, determine the significant features within the environmental change period based on the deviation vector, and extract specific data from the environmental feature set if the significant features exceed the preset threshold, and calculate the range of parameter change rate based on the specific data.

[0091] The mapping unit is used to obtain the parameter adjustment sequence under various operating conditions based on the parameter change rate range, map the parameter adjustment sequence with historical operating data, calculate the parameter mapping error between the new operating condition parameter set and the actual operating condition, and iteratively update the adaptive adjustment coefficients if the parameter mapping error exceeds the preset tolerance to obtain the optimized adjustment coefficient set.

[0092] The simulation unit is used to integrate the constraints of the adjustment coefficient set, verify the parameter consistency and boundary compliance under the constraints through simulation technology, adjust the weight distribution of the parameter combination and optimize to obtain the final control parameter combination.

[0093] The adaptive adjustment unit is used to apply the final control parameter combination to the servo motor system in real time, collect system feedback response data, analyze stability indicators, judge the system adaptability based on the stability indicator distribution, and generate operating status evaluation results.

[0094] This application also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the steps of the servo motor adaptive parameter adjustment method.

[0095] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0096] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the 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 cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for adaptive parameter adjustment of a servo motor, characterized in that, The method includes: The environmental data of the servo motor system during operation is collected by multiple sensors, the environmental data is fused into a comprehensive environmental vector, the comprehensive environmental vector is filtered for noise, and a smoothed environmental feature set is generated. Based on the environmental feature set, the deviation vector between the current working condition and the preset benchmark working condition is calculated. Based on the deviation vector, the significant features within the environmental change period are determined. If the significant features exceed the preset threshold, specific data is extracted from the environmental feature set, and the parameter change rate range is calculated based on the specific data. Based on the range of parameter change rates, obtain the parameter adjustment sequence under various working conditions, map the parameter adjustment sequence with historical operating data, calculate the parameter mapping error between the new working condition parameter set and the actual working condition, and if the parameter mapping error exceeds the preset tolerance, iteratively update the adaptive adjustment coefficient to obtain the optimized adjustment coefficient set. By integrating the constraints in the set of adjustment coefficients, the consistency of parameters and boundary compliance under the constraints are verified through simulation technology. The weight distribution of the parameter combination is adjusted and optimized to obtain the final combination of control parameters. The final control parameter combination is applied to the servo motor system in real time, system feedback response data is collected, and stability indicators are analyzed. Based on the distribution of the stability indicators, the system adaptability is judged and an operating status evaluation result is generated.

2. The servo motor adaptive parameter adjustment method according to claim 1, characterized in that, Generate a smoothed environmental feature set, including: The environmental data includes temperature data, humidity data, and voltage data. After normalizing the environmental data, they are combined to form a three-dimensional comprehensive environmental vector. A moving average filter is applied to smooth the data of each dimension of the comprehensive environmental vector to generate preliminary smooth environmental feature data. Fourier transform and wavelet transform are performed on the initially smoothed environmental feature data to identify abnormal fluctuation signals. The time-domain features and high-frequency components of the abnormal fluctuation signals are extracted to form a smoothed environmental feature set containing the impact frequency distribution features. The abnormal fluctuation signal is analyzed by time-frequency analysis using short-time Fourier transform to select key components and quantize them into vector form. The key component vectors are then fused into the smoothed environmental feature set.

3. The servo motor adaptive parameter adjustment method according to claim 1, characterized in that, Identify salient features within a period of environmental change, including: The deviation vector is obtained by performing an element-by-element subtraction operation between the standard parameter vector of the preset benchmark working condition and the current working condition vector corresponding to the smoothed environmental feature set. The components corresponding to the impact direction angle and impact duration are separated from the deviation vector, and the main frequency and amplitude peak of the components are extracted as feature values ​​by Fourier transform. The salient features within the period of environmental change are determined based on the standard deviation of the eigenvalues.

4. The servo motor adaptive parameter adjustment method according to claim 1, characterized in that, The range of the rate of change of the calculated parameters includes: Numerical sequences reflecting impact energy attenuation and impact contact area data including contact surface size and distribution density were selected from the smoothed environmental feature set as the specific data. Filter outliers in the specific data and arrange them into a discrete point set according to time series. Then, perform interpolation smoothing on the discrete point set in sequence to obtain a parameter change rate sequence. The maximum value in the parameter change rate sequence is identified as the upper limit of the parameter change rate, and the parameter change rate range is formed based on the upper limit of the parameter change rate.

5. The servo motor adaptive parameter adjustment method according to claim 1, characterized in that, Calculate the parameter mapping error, including: Based on historical operating data, the parameter similarity score between historical operating condition parameters and the current parameter change rate range is calculated. The influence ratio of different parameters under variable operating conditions is extracted as the parameter weight distribution. A mapping relationship is constructed based on the parameter similarity score and the parameter weight distribution. Based on the mapping relationship, the historical parameter sequence that best matches the current parameter change rate range is selected to form the parameter adjustment sequence under variable operating conditions. Identify the impact components in the parameter adjustment sequence that correspond to the impact direction deflection angle and impact duration, and linearly map the impact components to the new working condition coordinate system based on the similarity transformation matrix of the deviation vector, thereby generating a new set of working condition parameters that includes the impact force peak range and damping coefficient. The parameter mapping error is generated by comparing the differences between specific data in the new operating condition parameter set and the actual collected feedback response data using the root mean square error formula.

6. The servo motor adaptive parameter adjustment method according to claim 1, characterized in that, The optimized set of adjustment coefficients is obtained, including: When the parameter mapping error exceeds the preset tolerance range, an adaptive adjustment model is constructed based on the new operating condition parameter set. The adjustment coefficients in the adaptive adjustment model are iteratively adjusted using the gradient descent algorithm until the error is minimized. Calculate the variance of the adjustment coefficients after iteration. If the variance is less than the first preset value, the coefficients are determined to be converged. The converged adjustment coefficients are then summarized to form an initial set of adjustment coefficients. Test data is extracted to verify the applicability of the initial adjustment data coefficient set, and coefficient combinations are selected based on the applicability as the optimized adjustment coefficient set.

7. The servo motor adaptive parameter adjustment method according to claim 1, characterized in that, The final combination of control parameters obtained through optimization includes: The range of peak impact force and the interval of impact damping coefficient are selected from the set of adjustment coefficients as the constraint conditions. The average value of the peak impact force is calculated as the verification benchmark. The constraint conditions are then converted into vector form. A finite element simulation model is constructed, the vectors corresponding to the constraints are input into the finite element simulation model and a virtual load is applied, and multiple iterative simulations are run. The parameter consistency index is calculated by comparing the percentage deviation between the simulation output and the expected output, and the impact damping coefficient is checked to see if it is within the preset range. Based on the parameter consistency index and boundary violation rate in the simulation verification report, the weight adjustment factor is obtained through linear mapping, and the weight distribution of the parameter combination is updated based on the weight adjustment factor. Using the sum of squares of the deviations between the weight distribution and the validation results as the loss function, the gradient descent algorithm is used to optimize the weight distribution. Based on the loss function, the parameter values ​​corresponding to the minimum loss are selected to form the final control parameter combination.

8. The servo motor adaptive parameter adjustment method according to claim 1, characterized in that, Generate operational status assessment results, including: The system feedback response data is noise filtered using an average filtering method to extract the millisecond-level impact response delay from the occurrence of the impact to the system stabilization, and the historical comparison value of the current impact response delay and the historical average delay is calculated. The average value of the impact response delay is calculated, the variance of the historical comparison value of the parameter is calculated, and the average value and variance are calculated by weighted summation formula to obtain the comprehensive stability index. Based on the distribution of the comprehensive stability index, the adaptability of the servo motor system is determined, and the operating status evaluation result is generated.

9. A servo motor adaptive parameter adjustment system, used to implement the servo motor adaptive parameter adjustment method as described in any one of claims 1-8, characterized in that, The system includes: The acquisition unit is used to acquire environmental data during the operation of the servo motor system through multiple sensors, fuse the environmental data into a comprehensive environmental vector, filter noise from the comprehensive environmental vector, and generate a smoothed environmental feature set. The feature analysis unit is used to calculate the deviation vector between the current working condition and the preset benchmark working condition based on the environmental feature set, determine the significant features within the environmental change period based on the deviation vector, and if the significant features exceed the preset threshold, extract specific data from the environmental feature set and calculate the parameter change rate range based on the specific data. The mapping unit is used to obtain the parameter adjustment sequence under multiple working conditions according to the parameter change rate range, map the parameter adjustment sequence with historical operating data, calculate the parameter mapping error between the new working condition parameter set and the actual working condition, and if the parameter mapping error exceeds the preset tolerance, iteratively update the adaptive adjustment coefficient to obtain the optimized adjustment coefficient set. The simulation unit is used to integrate the constraints in the set of adjustment coefficients, verify the consistency of parameters and boundary compliance under the constraints through simulation technology, adjust the weight distribution of parameter combinations and optimize to obtain the final combination of control parameters. An adaptive adjustment unit is used to apply the final control parameter combination to the servo motor system in real time, collect system feedback response data, analyze stability indicators, judge the system adaptability based on the distribution of the stability indicators, and generate operating status evaluation results.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instruction is executed by the processor, it implements a servo motor adaptive parameter adjustment method as described in any one of claims 1-8.

Images