Servo driving speed control method and system based on position feedback

Abstract

The invention provides a servo driving speed control method and system based on position feedback, and relates to the technical field of industrial automation control, and the method comprises the steps: firstly collecting a real-time position feedback signal of a servo driving execution mechanism, and carrying out the analysis to obtain position feedback multi-dimensional feature information; a dynamic association model of position feedback and load change is established by fusing real-time load state signals, and a load position association mapping relation is obtained; pre-judging a position change trend according to the mapping relation and historical operation data and generating a predictive speed regulation reference; then dynamically iteratively correcting the reference based on the real-time position feedback signal to obtain a target speed adjusting instruction; and finally, an instruction is transmitted to an execution mechanism, and a real-time operation state signal is fed back to form closed-loop control. According to the invention, high-precision and high-stability servo driving speed control is realized.

Description

Technical Field

[0001] This invention relates to the field of industrial automation control technology, and more specifically, to a servo drive speed control method and system based on position feedback. Background Technology

[0002] In the field of industrial automation control, servo drive systems play a crucial role, widely used in various precision machining equipment, robots, and other applications. The accuracy and stability of their speed control directly affect the performance of the entire system and product quality. Currently, traditional servo drive speed control methods mainly suffer from the following problems.

[0003] On the one hand, some methods rely solely on preset speed commands for control, lacking real-time perception and feedback of the actual operating status of the servo drive actuator. When the load changes, the inability to adjust the speed in a timely manner can easily lead to problems such as speed fluctuations and inaccurate positioning, affecting machining accuracy and equipment stability.

[0004] On the other hand, while some methods introduce position feedback signals, they simply compare the current position with the target position and then adjust the speed based on the deviation. These control methods do not fully consider the temporal and spatial distribution characteristics of the position feedback signal, nor the impact of load changes on position and speed, making it difficult to achieve high-precision speed control. This is especially true in scenarios with frequent load changes or dynamic loads, where the control performance is even more unsatisfactory. Summary of the Invention

[0005] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a servo drive speed control method based on position feedback, the method comprising:

[0006] The real-time position feedback signal of the servo-driven actuator is collected, and the temporal and spatial distribution characteristics of the real-time position feedback signal are analyzed to obtain multi-dimensional feature information of the position feedback.

[0007] By integrating the real-time load status signal and position feedback multi-dimensional feature information of the servo-driven actuator, a dynamic correlation model between position feedback and load change is established to obtain the load position correlation mapping relationship.

[0008] Based on the load position correlation mapping relationship and the historical operating data of the servo drive actuator, the position change trend is predicted and a predictive speed adjustment benchmark is generated;

[0009] The predictive speed adjustment benchmark is dynamically iteratively corrected based on the real-time position feedback signal to obtain the dynamically corrected target speed adjustment command.

[0010] The target speed adjustment command is transmitted to the servo drive actuator to drive it to adjust its running speed. At the same time, the real-time running status signal of the servo drive actuator is fed back to the position feedback multi-dimensional feature acquisition and processing stage as the input signal for the next control cycle.

[0011] Furthermore, embodiments of the present invention also provide a servo drive speed control system based on position feedback, characterized in that it includes:

[0012] A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the above-described position feedback-based servo drive speed control method by executing the machine-executable instructions.

[0013] In another aspect, embodiments of the present invention also provide a computer program product, the computer program product including machine-executable instructions, the machine-executable instructions being stored in a computer-readable storage medium, the processor of the position feedback-based servo drive speed control system reading the machine-executable instructions from the computer-readable storage medium, the processor executing the machine-executable instructions, causing the position feedback-based servo drive speed control system to execute the aforementioned position feedback-based servo drive speed control method.

[0014] Based on the above, by collecting real-time position feedback signals from servo-driven actuators and deeply analyzing their temporal and spatial distribution characteristics, a dynamic correlation model is established by integrating real-time load status signals with multi-dimensional position feedback information. This yields a load-position correlation mapping relationship, enabling a better understanding of the load's impact on position and speed. Based on this load-position correlation mapping relationship and historical operating data, position change trends are predicted, and a predictive speed adjustment benchmark is generated. This achieves forward-looking planning for speed adjustment, effectively addressing the impact of load changes and making adjustments in advance, thus improving the system's response speed and stability. Dynamic iterative correction of the predictive speed adjustment benchmark based on real-time position feedback signals further ensures the accuracy and real-time performance of speed adjustment. The target speed adjustment command can be adjusted promptly according to actual operating conditions, ultimately transmitting the target speed adjustment command to the servo-driven actuator and feeding back real-time operating status signals to form a closed-loop control. This achieves high-precision, high-stability servo-driven speed control, effectively improving the performance and product quality of industrial automation systems. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the execution flow of the servo drive speed control method based on position feedback provided in an embodiment of the present invention.

[0016] Figure 2This is a schematic diagram of exemplary hardware and software components of a position feedback-based servo drive speed control system provided in an embodiment of the present invention. Detailed Implementation

[0017] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a position feedback-based servo drive speed control method according to an embodiment of the present invention. The following is a detailed description of the position feedback-based servo drive speed control method.

[0018] Step S110: Collect the real-time position feedback signal of the servo-driven actuator, analyze the temporal and spatial distribution characteristics of the real-time position feedback signal, and obtain multi-dimensional feature information of the position feedback.

[0019] Taking the servo drive system of an industrial robotic arm as an example, this robotic arm is used for precision parts assembly. The servo drive actuator includes a base rotary joint, an upper arm telescopic joint, a lower arm rotary joint, and an end effector translational joint. Each joint is equipped with a servo motor and position detection elements. When the robotic arm performs assembly tasks, the real-time position of each joint needs to be accurately collected and analyzed to achieve precise control of the running speed.

[0020] Step S111: Real-time position feedback signals are synchronously collected by multiple sets of position detection elements configured in the servo drive actuator. The multiple sets of position detection elements are evenly distributed in different moving parts of the servo drive actuator, and the collection range covers the entire motion trajectory of the actuator. The multiple sets of position detection elements synchronously start the collection program to capture the position change information of each part of the actuator.

[0021] In the robotic arm system, an absolute encoder is installed on the base rotary joint to detect the rotation angle position, a wire-type displacement sensor is installed on the upper arm telescopic joint to detect the telescopic length, an incremental encoder is installed on the forearm rotary joint to detect the rotation angle, and a grating ruler is installed on the end effector translation joint to detect the translation position in the X and Y directions. These four sets of position detection elements are distributed across different moving parts, covering the entire motion trajectory of the robotic arm from its initial position to its maximum working radius. When the robotic arm starts working, a synchronous trigger signal controls the four sets of position detection elements to simultaneously start the acquisition program, continuously capturing the position change information of each joint at 1-millisecond intervals. For example, the absolute encoder on the base rotary joint outputs the rotation angle value relative to the reference position in real time, the wire-type displacement sensor on the upper arm telescopic joint outputs the extension length value in real time, the incremental encoder on the forearm rotary joint calculates the rotation angle in real time through counting pulses, and the grating ruler on the end effector translation joint outputs the displacement in the X and Y directions respectively.

[0022] Step S112: Perform synchronous integration processing on the real-time position feedback signals collected by multiple sets of position detection elements, and associate and bind the position feedback signals of different moving parts according to the acquisition time dimension to generate a set of synchronous position feedback signals.

[0023] Each position detection element collects a real-time position feedback signal with its own acquisition timestamp. First, the timestamps are uniformly calibrated to ensure a consistent time base for all signals. Then, the position feedback signals of each moving part at the same moment are associated and bound according to their chronological order. For example, at acquisition time T1, the base rotation joint angle is A1, the upper arm extension length is L1, the forearm rotation angle is B1, and the end effector's X-direction displacement is X1 and Y-direction displacement is Y1. These data are combined into a tuple (T1, A1, L1, B1, X1, Y1). This process is repeated for each acquisition moment, ultimately forming a set of synchronous position feedback signals arranged chronologically. Each element in this set contains the position data of each moving part at the corresponding moment.

[0024] Step S113: Extract time-series signal data from the set of synchronous position feedback signals, perform time-series feature parsing processing, track the change trajectory of the real-time position feedback signal over time, extract the continuous change trend and periodic change pattern of the change trajectory, record the position information of key change nodes in the change trajectory, and generate position feedback time-series feature data.

[0025] The time-series signal data is selected from the synchronous position feedback signal set and sorted according to the chronological order of acquisition time to form a sequentially ordered time-series position feedback data. Trajectory tracking processing is performed on the ordered time-series position feedback data, connecting the position data at each time point in chronological order to form the trajectory of the real-time position feedback signal changing over time. Continuous trend analysis is performed on the changing trajectory, dividing it into time windows, analyzing the direction of trajectory change within each time window, and integrating the analysis results of each time window to obtain continuous trajectory change trend data. Periodic analysis is performed on the changing trajectory to find recurring position change patterns, determine the time interval of the recurrence patterns, extract the periodic change pattern of the trajectory, and generate periodic trajectory change data. Key node identification processing is performed on the changing trajectory to find nodes with abrupt changes in position change amplitude and nodes where the trajectory change trend changes, identifying key change nodes. The position information and corresponding time information of each key change node are recorded, establishing a correspondence between key change nodes and time, and generating key change node information. The system integrates and correlates continuous trajectory trend data, periodic trajectory change data, and key change node information. It establishes a correspondence between time-series features through timestamp association, creating a unified correspondence between each feature data point in the time dimension. The integrated time-series feature data undergoes feature normalization processing, extracting time-series features from the normalized data. Based on a preset variance threshold or importance score, it selects the main time-series features, integrates these features, and generates location feedback time-series feature data.

[0026] Step S1131: Select time-series signal data from the set of synchronous position feedback signals, sort the signal data according to the order of acquisition time, and form a time-series coherent and ordered time-series position feedback data.

[0027] Each element in the synchronous position feedback signal set contains a timestamp and position data for each moving part. Position data for each moving part at different time points is extracted as temporal dimension signal data. For example, the angle data sequence of the base rotary joint at each time point is extracted (T1, A1), (T2, A2), ..., (Tn, An). Similarly, the position data sequences of the upper arm telescopic joint, forearm rotary joint, and end effector in the X and Y directions are extracted. Then, the above data sequences are sorted according to the order of acquisition time to ensure that the position data of each moving part is arranged sequentially in time, forming ordered temporal position feedback data.

[0028] Step S1132: Perform trajectory tracking processing on the ordered temporal position feedback data, connect the position data of each time point in chronological order to form the trajectory of the real-time position feedback signal as time progresses.

[0029] Using time as the horizontal axis and the position data of each moving part as the vertical axis, each data point in the ordered temporal position feedback data is marked on the coordinate system. Then, adjacent data points are connected sequentially with line segments according to time order, forming a curve showing the change of position of each moving part over time, i.e., the trajectory of the real-time position feedback signal. For example, the trajectory of the base's rotating joint is a curve with time as the horizontal axis and angle as the vertical axis. This curve allows a direct visual observation of how the base's rotation angle changes over time.

[0030] Step S1133: Perform continuous trend analysis on the changing trajectory, divide the time window, analyze the direction of change of the trajectory within each time window, and integrate the analysis results of each time window to obtain the continuous trend data of trajectory change.

[0031] The time axis of the trajectory change is divided into multiple fixed-length time windows, for example, each time window is 100 milliseconds. For each time window, the difference between the starting position data and the ending position data within that window is calculated. If the difference is positive, the trajectory change direction within that window is considered positive; if the difference is negative, it is considered negative; if the difference is zero, it is considered unchanged. The change direction of each time window is recorded, and the analysis results of all time windows are integrated to form continuous trajectory change trend data. For example, if the base rotation angle increases from A0 to A1 within a certain time window, the difference is positive, and the change direction is considered positive; this result is recorded in the continuous trajectory change trend data.

[0032] Step S1134: Perform periodic analysis on the changing trajectory, find the recurring position change patterns in the trajectory, determine the time interval of the recurring patterns, extract the periodic change pattern of the trajectory, and generate periodic change data of the trajectory.

[0033] A sliding window method is used to scan the trajectory. A sliding window of a certain length is set, and the position change pattern within the window is compared with patterns at other positions to find recurring patterns. When a recurring pattern is found, the difference in the start time of two adjacent recurring patterns is calculated to determine the time interval of the recurrence. For example, if it is found that the base rotating joint exhibits a position change pattern of moving from angle A to angle B and back to angle A every 2 seconds, this 2-second time interval is the period in the periodic change pattern. This period and the corresponding position change pattern are recorded to generate trajectory periodic change data.

[0034] Step S1135: Perform key node identification processing on the changing trajectory, find nodes in the changing trajectory where the position change amplitude changes abruptly, find nodes where the trajectory change trend changes, and determine key change nodes.

[0035] Set thresholds for position change magnitude and trend change, and iterate through each data point in the change trajectory. For nodes with abrupt changes in position change magnitude, calculate the position difference between the current data point and the previous data point. If the absolute value of the difference is greater than the position change magnitude threshold, then the data point is a node with abrupt changes in position change magnitude. For nodes where the trajectory change trend changes, compare the change direction of the current time window with the change direction of the previous time window. If a change occurs (e.g., from positive to negative, from positive to unchanged, etc.), then the starting data point of that time window is a node where the trajectory change trend changes. Identify these two types of nodes as key change nodes.

[0036] Step S1136: Record the location information and corresponding time information of each key change node, establish the correspondence between key change nodes and time, and generate key change node information.

[0037] For each identified key change node, extract its corresponding location data and timestamp, and match the location data and timestamp one by one to form key change node information. For example, if the timestamp of a key change node is T5, and the location data is the base rotation angle A5, the boom extension length L5, etc., record the above information in the form of (T5, A5, L5, ...).

[0038] Step S1137: Link and integrate the trajectory continuous change trend data, trajectory periodic change data and key change node information, establish the correspondence between time series features through timestamp association, and establish a unified correspondence between each feature data in the time dimension.

[0039] Using timestamps as a link, data with the same timestamp or time interval in continuous trajectory trend data, periodic trajectory change data, and key change node information are associated. For example, in the time window containing timestamp T3, the continuous trajectory trend data shows a positive change, the periodic trajectory change data shows that the time window is within a certain period, and T3 is not a key node in the key change node information. By integrating the above information, a unified correspondence between them in the time dimension is established.

[0040] Step S1138: Perform feature normalization processing on the integrated time series feature data to extract time series features from the normalized time series feature data; based on a preset variance threshold or importance score, select the main time series features from the time series features; integrate the main time series features to generate location feedback time series feature data.

[0041] The integrated time-series feature data is normalized, transforming feature data of different magnitudes to the same data range, for example, normalizing location data to the range of 0-1. Then, time-series features such as trends, cycles, and key nodes are extracted from the normalized data. A preset variance threshold is used to calculate the variance of each time-series feature; features with variances greater than the threshold are selected as primary time-series features. Alternatively, importance scores for each time-series feature are obtained through model training, and features with scores higher than a set value are selected as primary time-series features. Finally, the selected primary time-series features are integrated to generate location feedback time-series feature data.

[0042] Step S114: Extract spatial dimension signal data from the set of synchronous position feedback signals, perform spatial distribution feature analysis processing, analyze the relative positional relationship and spatial distribution pattern between position feedback signals of different moving parts, explore the collaborative correlation characteristics of position changes of each part, and generate position feedback spatial distribution feature data.

[0043] Position data of different moving parts at the same moment are extracted from the synchronous position feedback signal set as spatial dimension signal data. For example, at time Tk, the base rotation angle Ak, the boom extension length Lk, the forearm rotation angle Bk, and the end effector X-direction displacement Xk and Y-direction displacement Yk are extracted. The position data of each moving part are transformed to a unified spatial coordinate system through coordinate transformation, and the relative distances and angular relationships between different moving parts are calculated. The changes of these relative positional relationships over time are analyzed to summarize spatial distribution patterns, such as the positional proportions between joints and the synergy of relative positional changes. The synergistic correlation characteristics of positional changes of each part are explored, such as how the position of the end effector changes as the boom extension length increases, and the degree of influence of the forearm rotation angle on the end effector position. The analyzed relative positional relationships, spatial distribution patterns, and synergistic correlation characteristics are integrated to generate spatial distribution characteristic data of position feedback.

[0044] Step S115: Normalize the temporal feature data and spatial distribution feature data of the location feedback respectively to generate normalized temporal feature data and normalized spatial feature data; perform correlation and fusion processing on the normalized temporal feature data and normalized spatial feature data to construct a temporal-spatial feature correlation matrix; and explore the inherent correlation between temporal changes and spatial distribution through correlation analysis to establish the correspondence between temporal changes and spatial distribution.

[0045] For location feedback time-series feature data, a min-max normalization method is used. This involves subtracting the minimum value of each feature from its actual value and then dividing by the difference between its maximum and minimum values ​​to obtain normalized time-series feature data. Similarly, the same normalization process is performed on the spatial distribution feature data of location feedback to generate normalized spatial feature data. The normalized time-series and spatial feature data are then arranged in chronological order to construct a two-dimensional matrix. The rows of the matrix represent the time series, and the columns represent different time-series and spatial features, forming the time-series-spatial feature correlation matrix. By calculating the correlation coefficients between different features in the matrix, the linear relationship between time-series and spatial features is analyzed to uncover inherent correlation patterns. For example, changes in a certain time-series feature will cause changes in which spatial features, and to what extent. This establishes a correspondence between time-series changes and spatial distribution.

[0046] Step S116: Extract core feature variables based on the temporal spatial feature correlation matrix. The core feature variables represent the key change information of the real-time position feedback signal and the spatial distribution differences of different parts. The core feature variables are the core indicators reflecting the position status of the actuator.

[0047] Calculate the variance contribution rate of each feature in the temporal spatial feature correlation matrix. The variance contribution rate represents the degree to which the feature contributes to the total variance of the data. Sort the features in descending order of variance contribution rate, and select the top few features whose cumulative variance contribution rate reaches a preset proportion (e.g., 85%) as core feature variables. These core feature variables can characterize the key change information of the real-time position feedback signal, such as the main temporal change trend and significant spatial distribution differences, and are the core indicators reflecting the position status of the actuator.

[0048] Step S117: Perform feature enhancement processing on the core feature variables to obtain enhanced core feature data; perform association mapping processing on the enhanced core feature data and the set of synchronous position feedback signals to establish the correspondence between the enhanced core feature data and the signals in the set of synchronous position feedback signals.

[0049] The core feature variables are enhanced by means of factors such as multiplying them by an enhancement coefficient to emphasize their importance or applying nonlinear transformations to improve their discriminative power, resulting in enhanced core feature data. Then, the enhanced core feature data is mapped to signals in the synchronous position feedback signal set based on timestamps, establishing a correspondence between each enhanced core feature data point and the original signal data at the corresponding time in the synchronous position feedback signal set.

[0050] Step S118: Integrate the enhanced core feature data, the temporal feature data of the location feedback, and the spatial distribution feature data of the location feedback to generate initial multi-dimensional feature information of the location feedback; perform dimensionality reduction and / or feature selection processing on the initial multi-dimensional feature information of the location feedback to generate multi-dimensional feature information of the location feedback.

[0051] The enhanced core feature data, temporal feature data, and spatial distribution feature data of location feedback are concatenated and integrated in chronological order to form initial multi-dimensional feature information of location feedback containing multiple feature dimensions. Principal component analysis is used to reduce the dimensionality of the initial multi-dimensional feature information of location feedback, mapping high-dimensional features to a low-dimensional space and retaining the main information; or feature selection is performed based on feature importance scores to remove unimportant features. Through dimensionality reduction and / or feature selection, the feature dimensions are reduced, generating the final multi-dimensional feature information of location feedback.

[0052] Step S120: Integrate the real-time load status signal and position feedback multi-dimensional feature information of the servo drive actuator to establish a dynamic correlation model between position feedback and load changes, and obtain the load position correlation mapping relationship.

[0053] In a robotic arm servo drive system, the real-time load status signal reflects the load borne by each joint during movement. By fusing it with multi-dimensional feature information from position feedback, a dynamic correlation between position and load can be established.

[0054] Step S121: The load detection unit configured in the servo drive actuator collects real-time load status signals. The load detection unit is deployed in the power transmission part of the servo drive actuator. The load detection unit collects load change information at different motion positions and different motion stages according to a preset collection cycle.

[0055] The load detection unit includes torque sensors mounted on the output shafts of each servo motor and force sensors mounted at the joints of the robotic arm. These load detection units are deployed in the power transmission area and can directly detect changes in load. The preset acquisition period is 1 millisecond, consistent with the acquisition period of the position detection elements, to ensure data time synchronization. During the robotic arm's movement, the load detection unit acquires load change information at different movement positions (such as the starting position, intermediate position, target position, etc.) and different movement stages (such as acceleration stage, constant speed stage, deceleration stage, etc.). For example, the torque sensors detect the torque value output by the servo motors, and the force sensors detect the magnitude and direction of the force at the joints.

[0056] Step S122: Perform signal purification processing on the real-time load status signal, extract load change features from the purified real-time load status signal, analyze the load change pattern over time to obtain the load change time sequence pattern, statistically analyze the range of load change amplitude to obtain the load change amplitude feature, record the duration of the load maintaining a specific state to obtain the load change duration feature, and integrate the load change time sequence pattern, load change amplitude feature and load change duration feature to generate a load change feature set.

[0057] The real-time load status signal is purified by using a low-pass filter to remove high-frequency noise and a median filter to eliminate pulse interference, resulting in a purified real-time load status signal. Load change characteristics, including the timing pattern of load changes, the amplitude characteristics of load changes, and the duration characteristics of load changes, are extracted from the purified real-time load status signal.

[0058] Step S1221: Sort the real-time load status signal after purification according to the time order to generate ordered load status data. The ordered load status data reflecting the continuous change of load over time is formed by sorting by time.

[0059] The purified real-time load status signals are arranged in chronological order according to the timestamps of the acquisition, forming ordered load status data. For example, the torque and force values ​​collected at each moment are arranged in chronological order to obtain a load data sequence that changes continuously over time.

[0060] Step S1222: Perform time-series pattern analysis on the ordered load state data, divide the data into fixed time intervals, analyze the direction and pattern of load change within each time interval, and integrate the analysis results of each time interval to obtain the time-series pattern of load change.

[0061] Divide the data into fixed time intervals, such as 200 milliseconds each, and analyze the direction (increase, decrease, or no change) and pattern (linear, non-linear, etc.) of load data changes within each time interval. For example, if the torque value increases linearly from T0 to T1 within a certain time interval, then the direction of change in that interval is positive, and the pattern is linear increase. Integrate the analysis results of all time intervals to obtain the temporal pattern of load changes.

[0062] Step S1223: Perform amplitude feature analysis on the ordered load state data, count the maximum and minimum values ​​of the load signal, determine the amplitude range of load changes, divide the amplitude intervals, calculate the distribution ratio of the load in different amplitude intervals, and generate load change amplitude features.

[0063] The maximum and minimum values ​​in the ordered load state data are statistically analyzed, and the difference between them represents the range of load variation. This range is then divided into multiple equally spaced intervals, such as dividing the 0-100N force range into intervals like 0-20N and 20-40N. The proportion of load data falling within each interval is calculated, representing the distribution ratio, thus generating the load variation amplitude characteristics.

[0064] Step S1224: Perform continuous feature analysis on the ordered load state data, identify the time periods during which the load maintains the same value or the same trend of change, record the duration of each time period, analyze the pattern of the load maintaining a specific state, and generate continuous load change features.

[0065] By comparing load data at adjacent time points, time periods in which the load maintains the same value (within a certain error range) or the same trend (such as continuous increase or continuous decrease) are identified. The start and end times of each time period are recorded, and the duration is calculated. The distribution and patterns of these durations are analyzed to generate a sustained characteristic of load changes. For example, if a uniform increase in load is observed over a period of 3-5 seconds, this duration information is part of the sustained characteristic.

[0066] Step S1225: Initially integrate the load change time sequence pattern, load change amplitude characteristics, and load change duration characteristics, establish the time correspondence between each feature through timestamp association, and establish the correspondence between each feature data and the same time interval.

[0067] Based on timestamps, characteristic data within the same time interval from the load change time sequence pattern, load change amplitude characteristics, and load change duration characteristics are correlated and integrated. For example, within the time interval T1-T2, the load change time sequence pattern is linearly increasing, the load change amplitude characteristic is in the range of 20-40N, and the load change duration characteristic is lasting for 3 seconds. These characteristic data are mapped to this time interval to establish a time correspondence.

[0068] Step S1226: Perform redundancy analysis on the initially integrated feature data, calculate the correlation coefficient between different features; delete duplicate features with correlation coefficients greater than a preset threshold, and retain features with correlation coefficients less than or equal to the preset threshold.

[0069] Calculate the correlation coefficients between the features after initial integration. The correlation coefficient measures the degree of linear correlation between two features. A preset correlation coefficient threshold, such as 0.8, is set. When the correlation coefficient between two features is greater than 0.8, they are considered to have strong redundancy, and one of the features is deleted. Features with a correlation coefficient less than or equal to 0.8 are retained to reduce feature dimensionality and redundant information.

[0070] Step S1227: Perform feature enhancement processing on the redundant feature data, reorder the enhanced feature data according to the time sequence of load changes, establish the temporal coherence relationship of the feature data, and generate ordered load change feature data.

[0071] Feature enhancement is performed on the redundant feature data, for example, by weighting or combining feature values ​​to enhance the expressive power of the features. Then, the enhanced feature data is reordered according to the time sequence of load changes to ensure that the feature data is coherent in the time dimension, generating ordered load change feature data.

[0072] Step S1228: Based on preset weights or contributions, extract the main load change features with weights or contributions higher than the threshold from the ordered load change feature data; integrate the core load change features and ordered load change feature data to generate a load change feature set.

[0073] Preset weight or contribution thresholds, and determine the weight or contribution of each feature through model training or expert experience. Extract features with weights or contributions higher than the threshold as the main load change features, and integrate these main load change features with the ordered load change feature data to form a load change feature set.

[0074] Step S123: Extract location feature data related to load changes from the multi-dimensional feature information of location feedback, filter out the time-series feature data and spatial distribution feature data of location feedback under different load conditions, eliminate location feature information unrelated to load changes through feature correlation analysis, and generate a load-related location feature set.

[0075] The correlation coefficients between each location feature in the multi-dimensional feature information of location feedback and each load feature in the load change feature set are calculated. Location features with correlation coefficients greater than a preset correlation threshold are identified as location feature data related to load changes. Based on different load states (e.g., light load, medium load, heavy load), corresponding location feedback time-series feature data and location feedback spatial distribution feature data are selected. Through feature correlation analysis, location feature information with weak correlation to load changes (correlation coefficients less than a preset threshold) is further eliminated, generating a load-related location feature set.

[0076] Step S124: Perform synchronous association processing on the load change feature set and the load associated location feature set according to the time dimension, match the load change data and location feature data within the same time interval, establish the time correspondence between the load change data and the location feature data through timestamp matching processing, and generate a load location synchronous association data group.

[0077] Data from the load change feature set and the load-related location feature set are matched according to timestamps to ensure that load change data and location feature data correspond within the same time interval. For example, load change feature data within timestamp T1-T2 is combined with location feature data within the same time interval into a data group containing load and location information within that time interval, generating a load-location synchronization association data group.

[0078] Step S125: Normalize the load change feature data and the corresponding location feature data in the load location synchronization association data group to generate normalized load change feature data and normalized location feature data; perform dynamic association modeling based on the normalized load change feature data and normalized location feature data to construct an association model that can characterize the dynamic mapping relationship between load change and location feedback. Use the normalized load change feature data as the model input and the corresponding normalized location feature data as the model output to establish the mapping relationship between the model input and the model output.

[0079] Using the same normalization method as in step S115, the load change feature data and location feature data in the load location synchronization correlation data group are normalized respectively to obtain normalized load change feature data and normalized location feature data. A neural network model is selected as the dynamic correlation model, which includes an input layer, a hidden layer, and an output layer. The number of neurons in the input layer is consistent with the dimension of the normalized load change feature data, and the number of neurons in the output layer is consistent with the dimension of the normalized location feature data. Multiple neurons are set in the hidden layer to capture nonlinear relationships. The normalized load change feature data is used as the model input, and the normalized location feature data is used as the model output. The weights and biases of the model are adjusted through training samples to establish a mapping relationship between the input and output.

[0080] Step S126: Perform training optimization processing on the dynamic correlation model using the historical operating data of the servo-driven actuator. Input the load position synchronous correlation data group in the historical operating data into the model and adjust the internal correlation parameters of the model to minimize the error between the model's predicted position feature data and the actual position feature data in the historical operating data.

[0081] Historical operational data recorded during the past operation of the servo drive actuator is collected, and load position synchronization correlation data sets are extracted from these data as training samples. The training samples are divided into training and validation sets. The normalized load change feature data from the training set is input into the dynamic correlation model to obtain the model's predicted position feature data. The mean squared error (MSE) between the predicted position feature data and the actual position feature data in the training set is calculated. The model's internal correlation parameters (such as weights and biases) are adjusted using the backpropagation algorithm to minimize the MSE. The model is validated using the validation set to evaluate its generalization ability. If the validation error is large, the model structure or training parameters are adjusted, and the training process is repeated until the model performance meets the requirements.

[0082] Step S127: Apply the trained and optimized dynamic association model to the newly collected load location synchronous association data group, input load change feature data to obtain the corresponding location feature prediction data, and generate load location association prediction data.

[0083] Once a new load location synchronization and correlation data set is collected, the load change characteristic data within it is normalized and then input into the trained and optimized dynamic correlation model. The model outputs corresponding location feature prediction data based on the established mapping relationship. This prediction data is then correlated with the corresponding timestamps to generate load location correlation prediction data.

[0084] Step S128: Perform comparative analysis on the load location association prediction data and the location feature data in the actual collected load location synchronous association data, and calculate the error between the prediction data and the actual data; based on the error, use the backpropagation algorithm to adjust the internal parameter weights of the dynamic association model.

[0085] The predicted location features from the load location correlation prediction data are compared point-by-point with the actual collected location feature data, and the absolute and relative errors for each data point are calculated. Overall error metrics, such as mean absolute error and root mean square error, are statistically analyzed. Based on these errors, the internal parameter weights of the dynamic correlation model are readjusted using the backpropagation algorithm to further reduce prediction errors and improve the model's prediction accuracy.

[0086] Step S129: Extract the core association rules between load changes and location feedback from the optimized dynamic association model, extract the location feedback feature change patterns corresponding to different load changes, and generate load location association mapping relationship.

[0087] The internal structure and parameters of the optimized dynamic correlation model are analyzed to extract the core correlation rules between load changes and position feedback. For example, when the load torque increases by 10%, the rate of change of the base rotation angle decreases by 5%; this rule is the core correlation rule. Simultaneously, the position feedback characteristic change patterns corresponding to different load changes (such as load increase, load decrease, and load stabilization) are extracted, such as the trend and magnitude of the position changes of each joint when the load increases. These rules and patterns are then integrated to generate a load-position correlation mapping relationship.

[0088] Step S130: Based on the load position correlation mapping relationship and the historical operating data of the servo drive actuator, predict the position change trend and generate a predictive speed adjustment benchmark.

[0089] By utilizing the load position correlation mapping relationship and historical operation data, it is possible to predict the future position change trend of the robotic arm, thereby establishing a corresponding speed adjustment benchmark.

[0090] Step S131: Collect historical operating data of the servo drive actuator. The historical operating data includes historical position feedback signals, historical load status signals, historical speed adjustment commands and corresponding operating status data. The data acquisition covers all design operating conditions and load conditions of the servo drive actuator.

[0091] Collect operational data of the robotic arm under different time periods and tasks, including historical position feedback signals of each joint, historical load status signals (such as torque and force), historical speed adjustment commands (such as motor speed commands), and corresponding operational status data (such as operational smoothness and position control accuracy). Ensure that data acquisition covers all designed operating conditions of the robotic arm, such as different motion trajectories, workloads (light, medium, and heavy loads), and operating speeds, to guarantee the comprehensiveness and representativeness of historical data.

[0092] Step S132: Perform data filtering processing on historical operating data, retain data that reflects normal operating status and typical load change conditions, generate a valid historical operating data set, extract historical position feedback feature data and historical load status data from the valid historical operating data set, combine the load position correlation mapping relationship, explore the correlation pattern between load change and position feedback and speed regulation under historical conditions, and generate historical correlation pattern data.

[0093] Historical operational data is filtered to remove abnormal data (such as erroneous data caused by sensor failure or data from abnormal equipment operation), retaining data reflecting normal operating conditions and typical load change conditions (such as sudden load increases or periodic load changes) to generate a valid historical operational data set. From this valid historical operational data set, historical position feedback characteristic data (such as temporal and spatial distribution characteristics) and historical load status data (such as load change characteristics) are extracted. Combined with the load-position correlation mapping relationship, the analysis examines how load changes affect position feedback under different historical operating conditions, and how position feedback and load changes jointly affect speed regulation commands, uncovering the correlation patterns and generating historical correlation pattern data.

[0094] Step S133: Extract the current position feedback feature data from the multi-dimensional feature information of the position feedback, combine it with the real-time load status signal after purification, establish the correspondence between the current data and the historical operating condition data through operating condition feature matching, determine the current operating condition and load status of the servo drive actuator, and generate the current operating status data.

[0095] The system extracts location feedback feature data from multi-dimensional feature information, including the current time and recent period, such as the current temporal trend and spatial distribution characteristics. Combined with the real-time load status signal after purification, it extracts the current load characteristics, such as load magnitude and trend. The current location feedback feature data and load characteristics are then matched with features from historical operating condition data to find the most similar historical operating conditions. This determines the current operating condition (e.g., a certain stage of assembly work) and load status (e.g., light load), generating current operating status data.

[0096] Step S134: Normalize the data in the current operating status data and the historical correlation pattern data respectively; perform matching analysis on the normalized current operating status data and the normalized historical correlation pattern data, find historical operating conditions similar to the current operating status through feature comparison, extract the position change trend and corresponding speed adjustment parameters under similar historical operating conditions, and generate similar operating condition reference data.

[0097] The feature data in the current operating status data and historical correlation data are normalized to ensure they fall within the same data range. Then, the similarity between the current operating status data and each historical operating condition data in the historical correlation data is calculated. Similarity can be calculated using methods such as cosine similarity and Euclidean distance. A similarity threshold is set, and historical operating conditions with similarities higher than the threshold are identified as similar to the current operating status. The position change trends (e.g., position change curves over time) and corresponding speed adjustment parameters (e.g., speed adjustment amplitude, adjustment frequency) of these similar historical operating conditions are extracted to generate reference data for similar operating conditions.

[0098] Step S1341: Extract core operating features from the current operating status data. The core operating features include the current location feedback core features, the current load status core features, and the current operating condition identification information. The extracted core operating features represent the current operating status.

[0099] The core operational features that characterize the current operational status are extracted from the current operational status data. The core features of the current position feedback may include the current position of each joint and the rate of change of position; the core features of the current load status may include the current load size and the rate of change of load; the current operating condition identification information may include the type of task being executed and the current operation stage. The above core operational features together describe the current operational status.

[0100] Step S1342: Extract the core feature data of each historical operating condition from the historical correlation pattern data. The core feature data of each historical operating condition includes the core features of historical location feedback, the core features of historical load status, and the identification information of historical operating conditions, and generate a set of core features of historical operating conditions.

[0101] From historical correlation data, for each historical operating condition, extract the corresponding historical position feedback core features (such as historical joint positions and position change rates), historical load status core features (such as historical load size and load change rate), and historical operating condition identification information (such as historical task type and operation stage). Combine the above data into a set of historical operating condition core features.

[0102] Step S1343: Normalize the core operating features of the current operating state and each historical core feature in the set of core features of historical operating conditions respectively; compare and correlate the core operating features of the current operating state after normalization with each historical core feature in the set of core features of historical operating conditions after normalization, and analyze the similarity between the core operating features of the current operating state and each historical core feature.

[0103] The min-max normalization method is used to normalize the core operating features of the current operating state and each historical core feature in the historical operating condition core feature set. Then, the cosine similarity is calculated to analyze the similarity between the current core operating features and each historical core feature. The closer the cosine similarity is to 1, the more similar the two are.

[0104] Step S1344: Compare the similarity between the core operating features of the current operating state and each historical core feature with the similarity matching threshold, filter out historical operating conditions whose similarity meets the matching threshold requirements, and determine historical operating conditions similar to the current operating state.

[0105] A preset similarity matching threshold, such as 0.85, is set. The cosine similarity between the current core operating features and each historical core feature is compared with this threshold. Historical operating conditions with similarity greater than or equal to the threshold are selected and identified as historical operating conditions similar to the current operating state.

[0106] Step S1345: Extract complete operating data corresponding to similar historical operating conditions from the effective historical operating data set. The complete operating data corresponding to similar historical operating conditions includes historical position feedback signals, historical load status signals, historical position change trend data, and corresponding historical speed adjustment parameters.

[0107] Based on the selected similar historical operating conditions, complete operating data corresponding to these conditions are extracted from the effective historical operating data set, including time series data of historical position feedback signals, time series data of historical load status signals, historical position change trend data (such as the direction, amplitude, and speed of position change), and corresponding historical speed adjustment parameters (such as speed command value, adjustment time, etc.).

[0108] Step S1346: Perform time alignment processing on the complete operating data of similar historical operating conditions. Using the time starting point of the current operating status data as the benchmark, adjust the time axis of the historical data to establish a time comparability relationship between the historical data and the current data.

[0109] Using the start time of the current operating status data as the base time point, the time axis of the complete operating data of similar historical operating conditions is shifted so that the start time of the historical data is aligned with the base time point. For example, if the current operating status starts at time T0 and a similar historical operating condition starts at time T0', then the time axis of that historical data is shifted as a whole (T0-T0'), so that the time of the historical data also starts from T0, thereby establishing a time comparability relationship between the historical data and the current data.

[0110] Step S1347: Extract position change trend data under similar historical operating conditions from the aligned historical operating data, analyze the direction, magnitude and speed of position change under historical operating conditions, and generate historical position change trend reference data.

[0111] From the historical operational data after time alignment, extract the change data of the position feedback signal over time, analyze the direction (e.g., positive or negative), magnitude (e.g., the amount of change), and speed (e.g., the amount of change per unit time) of the position change, and integrate the above information to generate historical position change trend reference data.

[0112] Step S1348: Extract speed adjustment parameters under similar historical operating conditions from the aligned historical operating data. The speed adjustment parameters under similar historical operating conditions include the direction, amplitude and adjustment frequency information of speed adjustment, and generate historical speed adjustment parameter reference data.

[0113] Extract relevant data of speed adjustment commands from the historical operation data after time alignment, analyze the direction (such as acceleration or deceleration), amplitude (such as the size of the adjustment amount), and adjustment frequency (such as the number of adjustments per unit time) of speed adjustment, and generate historical speed adjustment parameter reference data.

[0114] Step S1349: Associate and bind the historical position change trend reference data with the historical speed adjustment parameter reference data, and establish the correspondence between historical position changes and speed adjustment through timestamp matching; integrate the associated and bound historical position change trend reference data and historical speed adjustment parameter reference data to generate similar working condition reference data.

[0115] Historical position change trend reference data and historical speed adjustment parameter reference data are linked and bound based on timestamps, so that the position change trend at each point in time corresponds to the corresponding speed adjustment parameter. These linked data are then integrated to form similar operating condition reference data, which includes the correspondence between position changes and speed adjustments under similar historical operating conditions.

[0116] Step S135: Based on the load position association mapping relationship and similar working condition reference data, predict the future position change trend of the servo drive actuator, analyze the direction and magnitude of the future position change, and generate the position change trend prediction result.

[0117] By combining the load position correlation mapping relationship, the position change is predicted based on the current load change characteristics. At the same time, the historical position change trend in similar working condition reference data is referenced to comprehensively judge the future position change trend of the servo drive actuator. The direction of future position change (such as the base continuing to rotate, the boom continuing to extend, etc.) and the magnitude of change (such as the size of the rotation angle, the length of the extension, etc.) are analyzed to generate the position change trend prediction result.

[0118] Step S136: Determine the direction of speed adjustment based on the predicted position change trend, combine the speed adjustment parameters in the reference data of similar working conditions, set the range of speed adjustment, and generate the initial speed adjustment benchmark.

[0119] Based on the predicted position change trend, if the predicted position needs to be increased, the speed adjustment direction is to increase speed; if the predicted position needs to be decreased, the speed adjustment direction is to decrease speed. Combining historical speed adjustment parameters from similar operating condition reference data, the range of speed adjustment is determined, such as the maximum and minimum adjustment range during increase and decrease, thus generating an initial speed adjustment benchmark.

[0120] Step S137: Extract speed regulation effect data under similar operating conditions from the effective historical operation data set, analyze the position control accuracy data and operation stability data corresponding to different speed regulation benchmarks, and generate speed regulation effect evaluation data.

[0121] Extract position control accuracy data (such as the deviation between the actual position and the target position) and operational stability data (such as speed fluctuations and vibration amplitude) corresponding to different speed adjustment benchmarks under similar operating conditions from the effective historical operating data set. Analyze these data to evaluate the effectiveness of different speed adjustment benchmarks and generate speed adjustment effect evaluation data, such as a position control accuracy of ±0.01mm and good operational stability under a certain adjustment benchmark.

[0122] Step S138: Optimize and adjust the initial speed adjustment benchmark by combining the speed adjustment effect evaluation data, adjust the speed adjustment amplitude parameter according to the position control accuracy data, correct the speed adjustment transition parameter according to the operation stability data, and generate the optimized speed adjustment benchmark.

[0123] If the speed regulation effect evaluation data shows that the position control accuracy does not meet the requirements, the speed regulation amplitude parameter is adjusted according to the magnitude of the deviation, such as increasing or decreasing the adjustment amplitude. If the operational stability is poor, the speed regulation transition parameters are corrected, such as adjusting the magnitude of acceleration and deceleration to make the speed change more stable. Through these adjustments, an optimized speed regulation benchmark is generated.

[0124] Step S139: Based on the optimized speed adjustment benchmark and the predicted position change trend, construct the dynamic adjustment logic of the speed adjustment benchmark with the position change trend, and generate a predictive speed adjustment benchmark.

[0125] Based on the predicted position change trend, the appropriate speed adjustment strategy is determined for different position stages. For example, a larger adjustment range is used during periods of rapid position change, while a smaller adjustment range is used when approaching the target position. The optimized speed adjustment benchmark is then combined with these dynamic adjustment logics to generate a predictive speed adjustment benchmark that can dynamically adjust according to the position change trend.

[0126] Step S140: Based on the real-time position feedback signal, the predictive speed adjustment benchmark is dynamically iteratively corrected to obtain the dynamically corrected target speed adjustment command.

[0127] By comparing the real-time position feedback signal with the predictive speed adjustment reference, the reference is dynamically corrected to improve the accuracy of speed control.

[0128] Step S141: Continuously collect the real-time position feedback signal of the servo-driven actuator, and parse the real-time position feedback multi-dimensional feature information according to the steps of position feedback multi-dimensional feature acquisition and processing.

[0129] Using the same method as in step S110, the real-time position feedback signals of each joint of the robotic arm are continuously collected, and the multi-dimensional feature information of the real-time position feedback is analyzed to obtain the temporal features and spatial distribution features.

[0130] Step S142: Extract real-time position deviation feature data from the multi-dimensional feature information of real-time position feedback, compare the current position with the expected position corresponding to the predictive velocity adjustment benchmark, and obtain the feature information of position deviation.

[0131] Based on the predictive velocity adjustment benchmark, the expected position at the current moment is calculated. The current position data from the multi-dimensional feature information of real-time position feedback is compared with the expected position to calculate the position deviation (current position minus expected position). Feature information such as the magnitude and direction (positive or negative deviation) of the deviation is extracted to generate real-time position deviation feature data.

[0132] Step S143: Combine the real-time load status signal after purification to analyze the impact of load changes on the current position deviation. Establish the correlation between load changes and position deviation through load-deviation correlation analysis, determine whether the position deviation is caused by load changes, and generate load impact analysis results.

[0133] Analyze the changes in real-time load status signals after purification, such as whether the load suddenly increases or decreases. Through load-deviation correlation analysis, calculate the correlation coefficient between the load change and the position deviation. If the correlation coefficient is large, it is determined that the position deviation may be caused by the load change; otherwise, it may be caused by other factors, generating load impact analysis results.

[0134] Step S144: Based on the position deviation characteristic data and load influence analysis results, determine the correction direction of the predictive speed adjustment benchmark. If the position deviation is a positive deviation, set the correction direction to enhance the speed adjustment amplitude; if it is a negative deviation, set the correction direction to weaken the speed adjustment amplitude.

[0135] If the position deviation is positive (current position is greater than expected position), it indicates that the speed may be too high, and the speed adjustment range needs to be reduced. If the position deviation is negative (current position is less than expected position), it indicates that the speed may be too slow, and the speed adjustment range needs to be increased. Simultaneously, referring to the load impact analysis results, if the deviation is caused by an increase in load, a larger adjustment may be needed to generate a correction direction.

[0136] Step S145: Based on the magnitude of the position deviation, calculate the first correction component using the first correction function; based on the load influence analysis results, calculate the second correction component using the second correction function; according to the preset fusion rules, synthesize the first correction component and the second correction component to generate the speed adjustment reference correction amount.

[0137] The first correction function can be a linear function of the position deviation magnitude, such as the first correction component being equal to the position deviation magnitude multiplied by a proportionality coefficient. The second correction function can be determined based on the load change and correlation coefficient in the load impact analysis results, such as the second correction component being equal to the load change multiplied by the correlation coefficient and then multiplied by another proportionality coefficient. The preset fusion rule can be a weighted summation, where the first and second correction components are multiplied by their respective weights and then summed to obtain the speed regulation baseline correction.

[0138] Step S146: Apply the speed adjustment reference correction amount to the predictive speed adjustment reference and perform correction processing to obtain the corrected speed adjustment reference.

[0139] Add the speed adjustment reference correction amount to the predictive speed adjustment reference (if the correction amount is positive) or subtract it from the predictive speed adjustment reference (if the correction amount is negative), and perform the correction process to obtain the corrected speed adjustment reference.

[0140] Step S147: Apply the corrected speed adjustment reference to the simulated operating environment of the servo drive actuator, collect the position feedback signal after the simulated operation, and analyze the multi-dimensional feature information of the simulated position feedback.

[0141] A simulation environment for the servo-driven actuator is constructed, capable of simulating the kinematics and dynamics of the robotic arm. The corrected speed adjustment reference is input into the simulation environment to drive the simulated robotic arm movement. Position feedback signals are collected during the simulation and analyzed to obtain multi-dimensional feature information of the simulated position feedback according to the method in step S110.

[0142] Step S148: Compare the multi-dimensional feature information of the simulated position feedback with the expected position feature information, calculate the position deviation value, analyze the adjustment effect of the corrected speed adjustment benchmark, and generate correction effect evaluation data.

[0143] The position data from the simulated position feedback multi-dimensional feature information is compared with the expected position feature information (the target position set according to the task requirements), and the position deviation value (such as root mean square error) is calculated. Based on the deviation value, the adjustment effect of the corrected speed adjustment benchmark is analyzed, such as whether the deviation is within the allowable range, and the correction effect evaluation data is generated.

[0144] Step S149: When the correction effect evaluation data shows that the position deviation is greater than the preset deviation threshold, based on the multi-dimensional feature information of the simulated position feedback and the load status signal in the simulated operation, the correction amount calculation coefficient is reset, the speed adjustment benchmark correction amount is recalculated, the recalculated speed adjustment benchmark correction amount is applied to the corrected speed adjustment benchmark to perform a second correction process, and the second corrected speed adjustment benchmark is input into the simulated operation environment to repeat the simulation operation and effect evaluation steps, forming an iterative correction loop.

[0145] When the position deviation exceeds a preset deviation threshold (e.g., 0.02 mm), simulated position deviation feature data is extracted from the multi-dimensional feature information of the simulated position feedback to determine the magnitude and direction of the deviation. Load status signals during simulation are collected and purified, and load changes are analyzed. Based on the deviation and load changes, the proportional coefficients in the first and second correction functions are reset, and the speed adjustment reference correction is recalculated. The new correction is applied to the corrected speed adjustment reference to obtain a further corrected speed adjustment reference. This is then input into the simulation environment to repeat the simulation operation and effect evaluation steps until the position deviation is less than or equal to the deviation threshold.

[0146] For example, step S1491: When the correction effect evaluation data shows that the position deviation exceeds the reasonable range, extract the simulated position deviation feature data from the multi-dimensional feature information of the simulated position feedback, and determine the magnitude, direction and trend of the position deviation during the simulation operation.

[0147] When the position deviation exceeds the reasonable range (i.e., greater than the deviation threshold), the position deviation data of each joint is extracted from the multi-dimensional feature information of the simulated position feedback, and the magnitude (such as absolute value), direction (positive or negative), and trend of deviation change over time (such as whether the deviation is increasing or decreasing) of the deviation are calculated.

[0148] Step S1492: Collect the load status signal during the simulation operation, perform signal purification processing on the load status signal during the simulation operation, remove the interference components in the load status signal by filtering, generate the purified load status signal for the simulation operation, and analyze the load change information during the simulation operation.

[0149] During the simulation operation, load status signals are acquired and processed using the same signal purification methods as in step S122, such as low-pass filtering and median filtering, to remove interference components and generate a purified load status signal for the simulation operation. Analysis of this signal yields information on load magnitude, changing trends (increasing, decreasing, or remaining stable).

[0150] Step S1493: Combining the simulated position deviation characteristic data and the simulated load status signal after purification, analyze the reasons why the position deviation exceeds the reasonable range, and determine whether the correction range of the speed adjustment reference is insufficient or the correction direction is deviated.

[0151] If the load is increasing, but the position deviation is in the opposite direction (current position is less than expected), the correction may be insufficient, failing to provide a sufficient increase in speed. If the load is stable, but the position deviation is in the opposite direction to the expected one, the correction direction may be incorrect. A comprehensive analysis is needed to determine the cause of the deviation.

[0152] Step S1494: When it is determined that the correction range is insufficient, adjust the correction amount calculation coefficient based on the magnitude of the simulated position deviation, increase the correction amount calculation coefficient to improve the correction range, and recalculate the speed adjustment reference correction amount; when it is determined that there is a deviation in the correction direction, reverse the correction direction parameter, reset the correction coefficient, and calculate a new speed adjustment reference correction amount.

[0153] If the correction is deemed insufficient, the proportional coefficients in the first and second correction functions are increased proportionally according to the magnitude of the simulated position deviation, thereby increasing the calculated correction amount. If the correction direction is deemed to be in deviation, the correction direction parameter is changed from positive to negative or vice versa, the correction coefficient is reset, and a new speed adjustment reference correction amount is calculated.

[0154] Step S1495: Apply the recalculated speed adjustment reference correction amount to the current corrected speed adjustment reference, perform a second correction process, and obtain the second corrected speed adjustment reference.

[0155] The recalculated correction is then calculated (added or subtracted) from the current corrected speed adjustment reference to obtain the second corrected speed adjustment reference.

[0156] Step S1496: Input the speed adjustment reference after secondary correction into the simulation operation environment of the servo drive actuator, start a new round of simulation operation, and collect the position feedback signal after the simulation operation through the position detection simulation module.

[0157] The corrected speed adjustment benchmark is input into the simulation environment to start a new round of simulation, simulating the robotic arm moving according to the new speed benchmark. Position feedback signals during the movement are collected by the position detection simulation module.

[0158] Step S1497: Following the steps of acquiring and processing multi-dimensional features of position feedback, analyze the position feedback signals collected in the new round of simulation operation to obtain multi-dimensional feature information of the new round of simulation position feedback.

[0159] Following the method in step S110, the position feedback signal collected in the new round of simulation operation is analyzed to obtain multi-dimensional feature information of the new round of simulation position feedback.

[0160] Step S1498: Compare the multi-dimensional feature information of the new round of simulated position feedback with the expected position feature information, calculate the new round of position deviation value, and generate new round of correction effect evaluation data.

[0161] By comparing the multi-dimensional feature information of the new round of simulated position feedback with the expected position feature information, the position deviation value of the new round is calculated, and the evaluation data of the correction effect of the new round is generated.

[0162] Step S1499: Determine whether the positional deviation shown in the new round of correction effect evaluation data is within a reasonable range. If it still exceeds the range, repeat the steps of deviation cause analysis, correction amount calculation coefficient adjustment, speed adjustment benchmark correction amount recalculation, correction processing, simulation operation and effect evaluation until the positional deviation is within a reasonable range, forming an iterative correction cycle.

[0163] Determine whether the position deviation in the new round is less than or equal to the deviation threshold. If it still exceeds the threshold, repeat steps S1491 to S1498 until the position deviation is within a reasonable range, forming an iterative correction loop.

[0164] Step S1410: When the position deviation is less than or equal to the preset deviation threshold shown in the correction effect evaluation data in the iterative correction cycle, stop the iterative correction and use the corrected speed adjustment benchmark at this time as the target speed adjustment command.

[0165] When the evaluation data of the correction effect in a certain correction cycle shows that the position deviation is less than or equal to the preset deviation threshold, the iterative correction process is stopped, and the current corrected speed adjustment benchmark is determined as the target speed adjustment command.

[0166] Step S150: The target speed adjustment command is transmitted to the servo drive actuator to drive it to adjust its running speed. At the same time, the real-time running status signal of the servo drive actuator is fed back to the position feedback multi-dimensional feature acquisition and processing stage as the input signal for the next control cycle.

[0167] The generated target speed adjustment command is sent to the servo drive actuator to control the operating speed of the robotic arm and form a closed-loop feedback control.

[0168] Step S151: Convert the target speed adjustment command into a control signal format recognizable by the servo drive actuator, and generate a drive control signal through signal format conversion processing.

[0169] The target speed adjustment command is usually in the form of a digital signal, which needs to be converted into a signal format that the control module of the servo drive actuator can recognize, such as a pulse width modulation signal or an analog voltage signal, to generate a drive control signal.

[0170] Step S152: The drive control signal is transmitted to the control input terminal of the servo drive actuator through the signal transmission module, triggering the power adjustment unit of the servo drive actuator to adjust the running speed according to the drive control signal.

[0171] The signal transmission module uses an industrial bus (such as CAN bus or EtherCAT bus) to transmit drive control signals to the control input terminal of the servo drive actuator. After receiving the signal, the control input terminal triggers the power adjustment unit (such as a servo motor controller) to adjust the output speed of the servo motor according to the drive control signal, thereby adjusting the running speed of the robotic arm.

[0172] Step S153: During the process of adjusting the running speed of the servo drive actuator, real-time position feedback signals are continuously collected through multiple sets of position detection elements, and real-time load status signals are continuously collected through the load detection unit. The collection frequency is synchronized with the parsing frequency of the multi-dimensional feature acquisition and processing of position feedback.

[0173] During the process of adjusting the operating speed of the robotic arm, multiple sets of position detection elements and load detection units continuously collect real-time position feedback signals and real-time load status signals at a collection cycle of 1 millisecond. This ensures that the collection frequency is consistent with the resolution frequency of the multi-dimensional feature acquisition and processing of position feedback, thereby guaranteeing the real-time performance and synchronization of the data.

[0174] Step S154: The continuously collected real-time position feedback signal is transmitted to the position feedback multi-dimensional feature acquisition and processing stage as input data for a new round of position feedback multi-dimensional feature acquisition, and a new round of position feedback multi-dimensional feature parsing is performed.

[0175] The continuously collected real-time position feedback signals are transmitted to the position feedback multi-dimensional feature acquisition and processing stage in step S110, and used as input data for position feedback multi-dimensional feature analysis in the next control cycle. Steps S111 to S118 are then executed again to obtain a new round of position feedback multi-dimensional feature information.

[0176] Step S155: The continuously collected real-time load status signal is transmitted to the load dynamic association modeling processing stage as input data for a new round of load dynamic association modeling. The load location association mapping relationship is updated through the parameter update mechanism of the load dynamic association model.

[0177] The continuously collected real-time load status signals are transmitted to the load dynamic correlation modeling processing step in step S120 as input data for a new round of load dynamic correlation modeling. Using the new load location synchronous correlation data set, the model parameters are adjusted through the parameter update mechanism of the load dynamic correlation model (such as an online learning algorithm), and the load location correlation mapping relationship is updated.

[0178] Step S156: Based on the multi-dimensional feature information of the position feedback obtained in the new round of analysis and the updated load position association mapping relationship, perform a new round of predictive speed adjustment benchmark generation processing to obtain the updated predictive speed adjustment benchmark.

[0179] Using the new round of analysis to obtain multi-dimensional feature information of location feedback and the updated load location association mapping relationship, repeat steps S131 to S139 to generate an updated predictive speed adjustment benchmark.

[0180] Step S157: Based on the real-time position feedback signal collected in the new round, perform a new round of dynamic iterative correction processing on the updated predictive velocity adjustment benchmark to generate the updated target velocity adjustment command.

[0181] Based on the real-time position feedback signal collected in the new round, the updated predictive velocity adjustment benchmark is dynamically iteratively corrected according to the method of steps S141 to S1410 to generate the updated target velocity adjustment command.

[0182] Step S158: Convert the updated target speed adjustment command into a new drive control signal, and transmit it to the servo drive actuator through the signal transmission module to continuously adjust its running speed.

[0183] The updated target speed adjustment command is converted into a new drive control signal and transmitted to the servo drive actuator through the signal transmission module to continuously adjust the operating speed of the robotic arm to adapt to real-time position and load changes.

[0184] Step S159: Repeat the steps of real-time signal acquisition, multi-dimensional feature analysis, load correlation modeling, prediction benchmark generation, dynamic iterative correction and drive control signal transmission to form a closed-loop linkage control process of position, velocity and load through the signal feedback link.

[0185] By repeatedly executing steps S153 to S158, a closed-loop linkage control process of position, speed, and load is formed through a cyclical process of real-time signal acquisition, processing, modeling, prediction, correction, and control, thereby achieving precise speed control of the servo-driven actuator.

[0186] Step S1510: During the operation of the closed-loop linkage control process, the operating status of the servo drive actuator is continuously monitored. The speed, position and load data of the actuator are collected through the operating status monitoring module and used for the calculation of subsequent control cycles.

[0187] During the operation of the closed-loop linkage control process, the operation status monitoring module collects the speed, position and load data of the robotic arm in real time. This data is not only used for the calculation of the current control cycle, but also stored as part of the historical operation data.

[0188] In one exemplary embodiment, a position feedback-based servo drive speed control system is provided. This position feedback-based servo drive speed control system can be a terminal, server, etc., and its internal structure diagram can be as follows: Figure 2 As shown, the position feedback-based servo drive speed control system includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, near-field communication, or other technologies. When the computer program is executed by the processor, it implements a position feedback-based servo drive speed control method. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, or a button, trackball, or touchpad set on the housing of a position feedback-based servo drive speed control system, or an external keyboard, touchpad, or mouse, etc.

[0189] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A servo drive speed control method based on position feedback, characterized in that, The method includes: The real-time position feedback signal of the servo-driven actuator is collected, and the temporal and spatial distribution characteristics of the real-time position feedback signal are analyzed to obtain multi-dimensional feature information of the position feedback. By integrating the real-time load status signal and position feedback multi-dimensional feature information of the servo-driven actuator, a dynamic correlation model between position feedback and load change is established to obtain the load position correlation mapping relationship. Based on the load position correlation mapping relationship and the historical operating data of the servo drive actuator, the position change trend is predicted and a predictive speed adjustment benchmark is generated; The predictive speed adjustment benchmark is dynamically iteratively corrected based on the real-time position feedback signal to obtain the dynamically corrected target speed adjustment command. The target speed adjustment command is transmitted to the servo drive actuator to drive it to adjust its running speed. At the same time, the real-time running status signal of the servo drive actuator is fed back to the position feedback multi-dimensional feature acquisition and processing stage as the input signal for the next control cycle.

2. The servo drive speed control method based on position feedback according to claim 1, characterized in that, The process involves acquiring the real-time position feedback signal of the servo-driven actuator, analyzing its temporal and spatial distribution characteristics, and obtaining multi-dimensional feature information of the position feedback, including: The servo-driven actuator is equipped with multiple sets of position detection elements to synchronously collect real-time position feedback signals. The multiple sets of position detection elements are evenly distributed in different moving parts of the servo-driven actuator, and the collection range covers the entire motion trajectory of the actuator. The multiple sets of position detection elements synchronously start the collection program to capture the position change information of each part of the actuator. Synchronous integration processing is performed on the real-time position feedback signals collected by multiple sets of position detection elements. The position feedback signals of different moving parts are associated and bound according to the acquisition time dimension to generate a set of synchronous position feedback signals. Extract time-series signal data from the set of synchronous position feedback signals, perform time-series feature parsing processing, track the change trajectory of real-time position feedback signals over time, extract the continuous change trend and periodic change pattern of the change trajectory, record the position information of key change nodes in the change trajectory, and generate position feedback time-series feature data. Spatial dimension signal data is extracted from the set of synchronous position feedback signals, spatial distribution feature analysis is performed, the relative positional relationship and spatial distribution pattern between position feedback signals of different moving parts are analyzed, the collaborative correlation characteristics of position changes of each part are explored, and position feedback spatial distribution feature data are generated. The temporal feature data and spatial distribution feature data of location feedback are normalized respectively to generate normalized temporal feature data and normalized spatial feature data. The normalized temporal feature data and normalized spatial feature data are then correlated and fused to construct a temporal-spatial feature correlation matrix. Correlation analysis is used to explore the inherent correlation between temporal changes and spatial distribution and to establish the correspondence between temporal changes and spatial distribution. Core feature variables are extracted based on the temporal spatial feature correlation matrix. The core feature variables represent the key change information of the real-time position feedback signal and the spatial distribution differences of different parts. The core feature variables are the core indicators reflecting the position status of the actuator. Feature enhancement processing is performed on the core feature variables to obtain enhanced core feature data; the enhanced core feature data is then correlated and mapped with the set of synchronous position feedback signals to establish the correspondence between the enhanced core feature data and the signals in the set of synchronous position feedback signals. Integrate and enhance core feature data, temporal feature data of location feedback, and spatial distribution feature data of location feedback to generate initial multi-dimensional feature information of location feedback; perform dimensionality reduction and / or feature selection processing on the initial multi-dimensional feature information of location feedback to generate additional multi-dimensional feature information of location feedback.

3. The servo drive speed control method based on position feedback according to claim 1, characterized in that, The fusion of real-time load status signals and multi-dimensional feature information of position feedback from the servo drive actuator establishes a dynamic correlation model between position feedback and load changes, obtaining a load-position correlation mapping relationship, including: The load detection unit configured in the servo drive actuator collects real-time load status signals. The load detection unit is deployed in the power transmission part of the servo drive actuator. The load detection unit collects load change information at different motion positions and different motion stages according to a preset collection cycle. The real-time load status signal is cleaned up, load change features are extracted from the cleaned real-time load status signal, the load change time sequence pattern is obtained by analyzing the change pattern of load over time, the range of load change amplitude is obtained by statistically analyzing the range of load change amplitude, the duration of load maintaining a specific state is obtained by recording the duration of load change duration, and the load change time sequence pattern, load change amplitude feature and load change duration feature are integrated to generate a load change feature set. Extract location feature data related to load changes from multi-dimensional feature information of location feedback, filter out the time series feature data and spatial distribution feature data of location feedback under different load conditions, eliminate location feature information that is not related to load changes through feature correlation analysis, and generate a load-related location feature set. The load change feature set and the load-related location feature set are synchronized and associated according to the time dimension. The load change data and location feature data within the same time interval are matched. The time correspondence between the load change data and location feature data is established through timestamp matching, and a load location synchronized association data group is generated. The load change feature data and the corresponding location feature data in the load location synchronization correlation data group are normalized to generate normalized load change feature data and normalized location feature data. Based on the normalized load change feature data and normalized location feature data, dynamic correlation modeling is performed to construct a correlation model that can represent the dynamic mapping relationship between load change and location feedback. The normalized load change feature data is used as the model input and the corresponding normalized location feature data is used as the model output to establish the mapping relationship between the model input and the model output. The dynamic correlation model is trained and optimized by using historical operating data of the servo-driven actuator. The load position synchronous correlation data group in the historical operating data is input into the model, and the internal correlation parameters of the model are adjusted to minimize the error between the model's predicted position feature data and the actual position feature data in the historical operating data. The trained and optimized dynamic association model is applied to the newly collected load location synchronous association data set. The corresponding location feature prediction data is obtained by inputting load change feature data, and load location association prediction data is generated. Compare and analyze the location feature data in the load location correlation prediction data and the actual collected load location synchronous correlation data to calculate the error between the prediction data and the actual data; based on the error, use the backpropagation algorithm to adjust the internal parameter weights of the dynamic correlation model. The core association rules between load changes and location feedback are extracted from the optimized dynamic association model. The change patterns of location feedback characteristics corresponding to different load changes are extracted, and a load-location association mapping relationship is generated.

4. The servo drive speed control method based on position feedback according to claim 1, characterized in that, The step of predicting position change trends and generating predictive speed adjustment benchmarks based on the load position correlation mapping relationship and historical operating data of the servo drive actuator includes: Collect historical operating data of the servo drive actuator. The historical operating data includes historical position feedback signals, historical load status signals, historical speed adjustment commands and corresponding operating status data. The data acquisition covers all design operating conditions and load conditions of the servo drive actuator. Data filtering and processing are performed on historical operating data to retain data that reflects normal operating status and typical load change conditions, generating a valid historical operating data set. Historical location feedback feature data and historical load status data are extracted from the valid historical operating data set. Combined with the load location correlation mapping relationship, the correlation pattern between load change and location feedback and speed regulation under historical conditions is explored to generate historical correlation pattern data. Extract current position feedback feature data from multi-dimensional feature information of position feedback, combine it with real-time load status signal after purification, establish the correspondence between current data and historical operating condition data through operating condition feature matching, determine the current operating condition and load status of servo drive actuator, and generate current operating status data. Normalize the data in the current operating status data and the historical correlation data respectively; perform matching analysis on the normalized current operating status data and the normalized historical correlation data, find historical operating conditions similar to the current operating status through feature comparison, extract the position change trend and corresponding speed adjustment parameters under similar historical operating conditions, and generate similar operating condition reference data. Based on the load position correlation mapping relationship and similar working condition reference data, predict the future position change trend of the servo drive actuator, analyze the direction and magnitude of the future position change, and generate position change trend prediction results; Based on the predicted results of position change trends, the direction of speed adjustment is determined. Combined with the speed adjustment parameters in the reference data of similar working conditions, the range of speed adjustment is set, and an initial speed adjustment benchmark is generated. Extract speed regulation effect data under similar working conditions from the effective historical operation data set, analyze the position control accuracy data and operation stability data corresponding to different speed regulation benchmarks, and generate speed regulation effect evaluation data; The initial speed adjustment benchmark is optimized and adjusted based on the speed adjustment effect evaluation data, the speed adjustment amplitude parameter is adjusted based on the position control accuracy data, and the speed adjustment transition parameter is corrected based on the operation stability data to generate the optimized speed adjustment benchmark. Based on the optimized speed adjustment benchmark and the predicted position change trend, a dynamic adjustment logic for the speed adjustment benchmark as the position changes is constructed to generate a predictive speed adjustment benchmark.

5. The servo drive speed control method based on position feedback according to claim 1, characterized in that, The step of dynamically iteratively correcting the predictive speed adjustment benchmark based on real-time position feedback signals to obtain a dynamically corrected target speed adjustment command includes: The real-time position feedback signal of the servo-driven actuator is continuously collected, and the real-time position feedback multi-dimensional feature information is obtained by parsing according to the steps of position feedback multi-dimensional feature acquisition and processing. Real-time position deviation feature data is extracted from multi-dimensional feature information of real-time position feedback, and the current position is compared with the expected position corresponding to the predictive speed adjustment benchmark to obtain the feature information of position deviation. By combining the real-time load status signal after purification, the impact of load changes on the current position deviation is analyzed. The correlation between load changes and position deviation is established through load-deviation correlation analysis to determine whether the position deviation is caused by load changes and generate load impact analysis results. Based on the position deviation characteristic data and load influence analysis results, the correction direction of the predictive speed adjustment benchmark is determined. If the position deviation is positive, the correction direction is set to enhance the speed adjustment amplitude; if it is negative, the correction direction is set to weaken the speed adjustment amplitude. Based on the magnitude of the position deviation, a first correction component is calculated using a first correction function; based on the load influence analysis results, a second correction component is calculated using a second correction function; according to a preset fusion rule, the first correction component and the second correction component are synthesized to generate a speed adjustment reference correction amount. The speed adjustment reference correction is applied to the predictive speed adjustment reference, and the correction process is performed to obtain the corrected speed adjustment reference. The corrected speed adjustment benchmark is applied to the simulated operating environment of the servo drive actuator, the position feedback signal after the simulated operation is collected, and the multi-dimensional feature information of the simulated position feedback is obtained by analysis; By comparing the multi-dimensional feature information of the simulated position feedback with the feature information of the expected position, the position deviation value is calculated, the adjustment effect of the corrected speed adjustment benchmark is analyzed, and the correction effect evaluation data is generated. When the correction effect evaluation data shows that the position deviation is greater than the preset deviation threshold, the correction amount calculation coefficient is reset based on the multi-dimensional feature information of the simulated position feedback and the load status signal in the simulated operation. The speed adjustment benchmark correction amount is recalculated, and the recalculated speed adjustment benchmark correction amount is applied to the corrected speed adjustment benchmark to perform another correction process. The corrected speed adjustment benchmark is then input into the simulated operation environment to repeat the simulation operation and effect evaluation steps, forming an iterative correction cycle. When the position deviation is less than or equal to the preset deviation threshold as the evaluation data of the correction effect in the iterative correction cycle is displayed, the iterative correction is stopped, and the corrected speed adjustment benchmark at this time is used as the target speed adjustment command.

6. The servo drive speed control method based on position feedback according to claim 2, characterized in that, The process involves extracting time-series signal data from the synchronous position feedback signal set, performing time-series feature parsing processing, tracking the trajectory of the real-time position feedback signal over time, extracting the continuous trend and periodic variation of the trajectory, recording the position information of key change nodes in the trajectory, and generating position feedback time-series feature data, including: The time-series signal data is selected from the set of synchronous position feedback signals and sorted according to the order of acquisition time to form a time-series coherent and ordered time-series position feedback data. Perform trajectory tracking processing on ordered temporal position feedback data, connect the position data of each time point in chronological order to form the trajectory of the real-time position feedback signal as time progresses; Continuous trend analysis is performed on the changing trajectory, time windows are divided, the direction of trajectory change within each time window is analyzed, and the analysis results of each time window are integrated to obtain continuous trajectory change trend data. Perform periodic analysis on the changing trajectory, find the recurring position change patterns in the trajectory, determine the time interval of the recurrence pattern, extract the periodic change pattern of the trajectory, and generate periodic change data of the trajectory. Perform key node identification processing on the changing trajectory, find nodes in the changing trajectory where the position change amplitude changes abruptly, find nodes where the trajectory change trend changes, and determine key change nodes; Record the location information and corresponding time information of each key change node, establish the correspondence between key change nodes and time, and generate key change node information; The data on continuous trajectory change trends, periodic trajectory change data, and key change node information are linked and integrated. The correspondence between time series features is established through timestamp association, and a unified correspondence between various feature data in the time dimension is established. The integrated time-series feature data is subjected to feature normalization processing to extract time-series features from the normalized time-series feature data; based on a preset variance threshold or importance score, the main time-series features are selected from the time-series features; the main time-series features are integrated to generate location feedback time-series feature data.

7. The servo drive speed control method based on position feedback according to claim 3, characterized in that, The process involves extracting load change features from the real-time load status signal after purification, analyzing the load change pattern over time to obtain the load change time series pattern, statistically analyzing the range of load change amplitude to obtain the load change amplitude feature, recording the duration of the load maintaining a specific state to obtain the load change duration feature, and integrating the load change time series pattern, load change amplitude feature, and load change duration feature to generate a load change feature set, including: The real-time load status signals after purification are sorted in chronological order to generate ordered load status data. The ordered load status data reflecting the continuous change of load over time is formed by sorting by time. Perform time-series pattern analysis on ordered load state data, divide fixed time intervals, analyze the direction and pattern of load change within each time interval, and integrate the analysis results of each time interval to obtain the time-series pattern of load change. Amplitude characteristic analysis is performed on ordered load state data to statistically analyze the maximum and minimum values ​​of load signals, determine the amplitude range of load changes, divide amplitude intervals, calculate the distribution ratio of load in different amplitude intervals, and generate load change amplitude characteristics. Perform continuous feature analysis on ordered load state data to identify time periods in which the load maintains the same value or the same trend of change, record the duration of each time period, analyze the pattern of the load maintaining a specific state, and generate continuous load change features. The timing pattern of load changes, the magnitude of load changes, and the duration of load changes are initially integrated. The time correspondence between each feature is established by timestamp association, and the correspondence between each feature data and the same time interval is established. Redundancy analysis is performed on the initially integrated feature data to calculate the correlation coefficient between different features; duplicate features with correlation coefficients greater than a preset threshold are deleted, and features with correlation coefficients less than or equal to the preset threshold are retained. The feature data after redundancy processing is subjected to feature enhancement processing. The enhanced feature data is then reordered according to the time sequence of load changes to establish a temporal coherence relationship of the feature data and generate ordered load change feature data. Based on preset weights or contribution levels, extract the main load change features with weights or contribution levels higher than a threshold from ordered load change feature data; Integrate core load change characteristics and ordered load change characteristics data to generate a load change characteristic set.

8. The servo drive speed control method based on position feedback according to claim 4, characterized in that, The data in the current operating status data and the historical correlation data are normalized respectively; the normalized current operating status data and the normalized historical correlation data are matched and analyzed, and historical operating conditions similar to the current operating status are found by feature comparison. The position change trend and corresponding speed adjustment parameters under similar historical operating conditions are extracted to generate similar operating condition reference data, including: Extract core operational features from the current operational status data. These core operational features include the current location feedback core features, the current load status core features, and the current operational condition identification information. The extracted core operational features characterize the current operational status. Extract the core feature data of each historical operating condition from historical correlation data. The core feature data of each historical operating condition includes the core features of historical location feedback, the core features of historical load status, and the identification information of historical operating conditions, and generate a set of core features of historical operating conditions. The core operating features of the current operating state and each historical core feature in the set of core features of historical operating conditions are normalized respectively; the core operating features of the current operating state after normalization are correlated and compared with each historical core feature in the set of core features of historical operating conditions after normalization, and the similarity between the core operating features of the current operating state and each historical core feature is analyzed. The similarity between the core operating features of the current operating status and each historical core feature is compared with the similarity matching threshold. Historical operating conditions that meet the similarity matching threshold requirements are selected, and historical operating conditions similar to the current operating status are identified. Extract complete operating data corresponding to similar historical operating conditions from the effective historical operating data set. The complete operating data corresponding to similar historical operating conditions includes historical position feedback signals, historical load status signals, historical position change trend data, and corresponding historical speed adjustment parameters. Perform time-series alignment processing on complete operating data of similar historical operating conditions, and adjust the time axis of historical data based on the time starting point of the current operating status data to establish a time comparability relationship between historical data and current data; Extract position change trend data under similar historical operating conditions from the aligned historical operating data, analyze the direction, magnitude and speed of position change under historical operating conditions, and generate historical position change trend reference data; Extract speed adjustment parameters under similar historical operating conditions from the aligned historical operating data. The speed adjustment parameters under similar historical operating conditions include the direction, amplitude and frequency of speed adjustment, and generate historical speed adjustment parameter reference data. The historical position change trend reference data is linked and bound with the historical speed adjustment parameter reference data, and the correspondence between historical position changes and speed adjustment is established by matching timestamps. By integrating and binding historical location change trend reference data and historical speed adjustment parameter reference data, similar working condition reference data is generated.

9. The servo drive speed control method based on position feedback according to claim 1, characterized in that, The process of transmitting the target speed adjustment command to the servo drive actuator to drive it to adjust its operating speed, and simultaneously feeding back the real-time operating status signal of the servo drive actuator to the position feedback multi-dimensional feature acquisition and processing stage as the input signal for the next control cycle, includes: The target speed adjustment command is converted into a control signal format that can be recognized by the servo drive actuator, and the drive control signal is generated through signal format conversion processing. The drive control signal is transmitted to the control input terminal of the servo drive actuator through the signal transmission module, which triggers the power adjustment unit of the servo drive actuator to adjust the running speed according to the drive control signal. During the process of adjusting the running speed of the servo drive actuator, real-time position feedback signals are continuously collected through multiple sets of position detection elements, and real-time load status signals are continuously collected through the load detection unit. The collection frequency is synchronized with the parsing frequency of the multi-dimensional feature acquisition and processing of position feedback. The continuously collected real-time location feedback signals are transmitted to the location feedback multi-dimensional feature acquisition and processing stage, serving as input data for a new round of location feedback multi-dimensional feature acquisition, and a new round of location feedback multi-dimensional feature analysis is performed. The continuously collected real-time load status signals are transmitted to the load dynamic correlation modeling processing stage as input data for a new round of load dynamic correlation modeling. The load location correlation mapping relationship is updated through the parameter update mechanism of the load dynamic correlation model. Based on the multi-dimensional feature information of the position feedback obtained from the new round of analysis and the updated load position association mapping relationship, a new round of predictive speed adjustment benchmark generation processing is performed to obtain the updated predictive speed adjustment benchmark. Based on the newly acquired real-time position feedback signal, a new round of dynamic iterative correction processing is performed on the updated predictive velocity adjustment benchmark to generate an updated target velocity adjustment command; The updated target speed adjustment command is converted into a new drive control signal and transmitted to the servo drive actuator through the signal transmission module to continuously adjust its running speed. The process involves repeatedly performing steps such as real-time signal acquisition, multi-dimensional feature analysis, load correlation modeling, prediction benchmark generation, dynamic iterative correction, and drive control signal transmission, forming a closed-loop linkage control process for position, velocity, and load through a signal feedback link. During the operation of the closed-loop linkage control process, the operating status of the servo drive actuator is continuously monitored. The speed, position and load data of the actuator are collected through the operating status monitoring module and used for the calculation of subsequent control cycles.

10. A servo drive speed control system based on position feedback, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the position feedback-based servo drive speed control method according to any one of claims 1 to 9 by executing the machine-executable instructions.

Images