Mechanical arm trajectory tracking control method and system

Abstract

The invention relates to the technical field of mechanical arm trajectory tracking control, in particular to a mechanical arm trajectory tracking control method and system. The method comprises the following steps that a mechanical arm operation log and a track execution program are obtained, periodic track motion deviation data are extracted, and track jitter sensing record analysis is carried out; thirdly, track behavior deviation feature learning is carried out according to the track deviation data, then a track tracking execution program is optimized, and the accuracy of track tracking control is improved; and finally, designing trajectory tracking control firmware, and embedding the trajectory tracking control firmware into a mechanical arm control center to realize optimized mechanical arm trajectory tracking control. The mechanical arm trajectory tracking control technology is optimized, so that the mechanical arm trajectory tracking control technology is more perfect.

Description

Technical Field

[0001] The present invention relates to the technical field of robotic arm trajectory tracking control, and in particular, to a method and system for robotic arm trajectory tracking control. Background Art

[0002] Robotic arms are increasingly widely used in various fields such as production, assembly, and handling. The trajectory tracking control technology of robotic arms is one of its core technologies, which directly affects the accuracy, efficiency, and stability of robotic arms in actual operations. Trajectory tracking control requires the robotic arm to accurately and smoothly follow a predetermined trajectory for operation. Especially when facing complex environments and dynamic changes, how to maintain the accurate execution of the trajectory and avoid trajectory deviations caused by error accumulation or external interference is the key to achieving efficient operation. However, when the robotic arm wears and ages, there will be a certain deviation between the actual movement trajectory of the robotic arm and the movement trajectory set by manual programming. A traditional method for robotic arm trajectory tracking control cannot accurately learn the deviation caused by aging and wear of the robotic arm after a certain usage period, resulting in a large problem of robotic arm trajectory tracking control error. Summary of the Invention

[0003] Based on this, it is necessary to provide a method and system for robotic arm trajectory tracking control to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for robotic arm trajectory tracking control, the method includes the following steps: Step S1: Obtain the operation log of the robotic arm and the robotic arm trajectory tracking execution program; extract the trajectory motion deviation associated with the usage period from the operation log of the robotic arm to obtain the periodic trajectory motion deviation data; perform analysis on the perception record of the robotic arm motion trajectory jitter based on the periodic trajectory motion deviation data to obtain the trajectory deviation jitter perception record data; Step S2: Perform learning on the trajectory behavior deviation characteristics of the robotic arm according to the trajectory deviation jitter perception record data to obtain the trajectory behavior progressive deviation learning data; perform iterative optimization on the robotic arm trajectory tracking execution program according to the trajectory behavior progressive deviation learning data to obtain the trajectory tracking execution optimization program; Step S3: Design the robotic arm trajectory tracking control firmware for the trajectory tracking execution optimization program to obtain the robotic arm trajectory tracking control firmware; embed the robotic arm trajectory tracking control firmware into the robotic arm control center to perform robotic arm trajectory tracking control.

[0005] Preferably, step S1 includes the following steps: Step S11: Obtain the operation log of the robotic arm and the robotic arm trajectory tracking execution code; Step S12: Clean the data of the operation log of the robotic arm to obtain the cleaned operation log of the robotic arm; Step S13: Extract the trajectory motion deviation associated with the usage cycle from the cleaning log of the robotic arm operation to obtain the periodic trajectory motion deviation data; Step S14: Based on the periodic trajectory motion deviation data, perform analysis on the perception record of the robotic arm motion trajectory jitter in the cleaning log of the robotic arm operation to obtain the trajectory deviation jitter perception record data.

[0006] Preferably, step S2 includes the following steps: Step S21: Analyze the skewness intensity of the trajectory jitter amplitude change in the trajectory deviation jitter perception record data to obtain the trajectory jitter amplitude skewness intensity data; Step S22: Conduct progressive analysis of the spatial deviation amount on the periodic trajectory motion deviation data to obtain the progressive data of the deviation amount between trajectory motions; Step S23: Based on the trajectory jitter amplitude skewness intensity data and the progressive data of the deviation amount between trajectory motions, perform learning on the progressive deviation characteristics of the robotic arm trajectory behavior to obtain the progressive deviation learning data of the trajectory behavior; Step S24: According to the progressive deviation learning data of the trajectory behavior, perform iterative optimization on the trajectory tracking execution program of the robotic arm to obtain the optimized trajectory tracking execution program.

[0007] Preferably, step S21 includes the following steps: Step S211: Perform time-domain to frequency-domain conversion on the trajectory deviation jitter perception record data to generate a deviation jitter time-frequency domain diagram; Step S212: Identify the skewness of the distribution of jitter frequency mutation points in the deviation jitter time-frequency domain diagram to obtain the skewness data of the distribution of jitter frequency mutation points; Step S213: Calculate the variance of the increasing slope of the mutation points based on the skewness data of the distribution of jitter frequency mutation points to obtain the variance of the increasing slope of the frequency mutation points; Step S214: Based on the variance of the increasing slope of the frequency mutation points, perform convergence constraint logarithmic fitting processing on the deviation jitter time-frequency domain diagram to obtain the convergence logarithmic fitting data of the frequency mutation points; Step S215: Analyze the skewness intensity of the trajectory jitter amplitude change according to the convergence logarithmic fitting data of the frequency mutation points to obtain the trajectory jitter amplitude skewness intensity data.

[0008] Preferably, step S214 includes the following steps: Based on the variance of the increasing slope of the frequency mutation points, perform incremental piecewise exponential transformation processing on the frequency domain curve of the mutation points in the deviation jitter time-frequency domain diagram to obtain the incremental piecewise exponential of the mutation point curve; Perform convergence processing on the ratio of the incremental piecewise exponential of the mutation point curve to obtain the convergence ratio of the incremental piecewise exponential; Perform a convergence constraint logarithmic fitting process on the deviation jitter time-frequency domain diagram according to the increasing segmented exponential convergence ratio to obtain the convergence logarithmic fitting data of the frequency mutation point.

[0009] Preferably, step S23 includes the following steps: Step S231: Calculate the progressive mean difference of the attitude angle offset for the progressive data of the deviation amount between trajectory movements to obtain the progressive mean difference of the attitude angle offset; Step S232: Calculate the ratio of the time-sequence increment of the joint angle deviation for the progressive data of the deviation amount between trajectory movements to obtain the ratio of the time-sequence increment of the joint angle deviation; Step S233: Perform a regression analysis of the mechanical arm trajectory jitter swing amplitude on the progressive mean difference of the attitude angle offset and the ratio of the time-sequence increment of the joint angle deviation according to the trajectory jitter amplitude skewness intensity data to obtain the regression data of the mechanical arm trajectory jitter swing amplitude; Step S234: Perform transfer learning of the repeated trajectory jitter swing amplitude on the regression data of the mechanical arm trajectory jitter swing amplitude to obtain the learning data of the repeated trajectory jitter swing amplitude; Step S235: Perform learning on the progressive deviation characteristics of the mechanical arm trajectory behavior according to the learning data of the repeated trajectory jitter swing amplitude to obtain the learning data of the progressive deviation of the trajectory behavior.

[0010] Preferably, step S233 includes the following steps: Analyze the disordered intensity of the jitter space azimuth for the trajectory jitter amplitude skewness intensity data to obtain the disordered intensity data of the jitter space azimuth; Perform an analysis of the mutation entropy value of the azimuth vector angular rate on the progressive mean difference of the attitude angle offset according to the disordered intensity data of the jitter space azimuth to obtain the mutation entropy value of the azimuth vector angular rate; Perform a calculation of the joint angle overshoot probability on the ratio of the time-sequence increment of the joint angle deviation according to the disordered intensity data of the jitter space azimuth to obtain the calculation data of the joint angle overshoot probability; Perform a regression analysis of the mechanical arm trajectory jitter swing amplitude based on the mutation entropy value of the azimuth vector angular rate and the calculation data of the joint angle overshoot probability to obtain the regression data of the mechanical arm trajectory jitter swing amplitude.

[0011] Preferably, step S24 includes the following steps: Step S241: Perform a clustering process on the mechanical arm trajectory error for the learning data of the progressive deviation of the trajectory behavior to obtain the clustering data of the mechanical arm trajectory error; Step S242: Reconstruct the control logic structure of the mechanical arm trajectory tracking execution program based on the clustering data of the mechanical arm trajectory error to generate the reconstructed data of the control program logic structure; Step S243: Constrain the nested depth of the control logic for the reconstructed data of the control program logic structure to obtain the reconstructed data of the depth constraint of the program control logic. Step S244: Reconstruct the data according to the depth constraint of the program control logic, and perform iterative optimization on the trajectory tracking execution program of the robotic arm to obtain an optimized trajectory tracking execution program.

[0012] Preferably, step S3 includes the following steps: Step S31: Perform static analysis on the program code of the optimized trajectory tracking execution program to obtain static analysis data of the program code; Step S32: Perform integrated testing of functional units on the optimized trajectory tracking execution program according to the static analysis data of the program code to obtain integrated testing data of functional units; Step S33: Design the robotic arm trajectory tracking control firmware based on the integrated testing data of functional units for the optimized trajectory tracking execution program to obtain the robotic arm trajectory tracking control firmware; Step S34: Embed the robotic arm trajectory tracking control firmware into the robotic arm control center to perform robotic arm trajectory tracking control.

[0013] Preferably, the present invention also provides a robotic arm trajectory tracking control system for executing the above-mentioned robotic arm trajectory tracking control method. The robotic arm trajectory tracking control system includes: A trajectory jitter perception and analysis module, configured to obtain the robotic arm operation log and the robotic arm trajectory tracking execution program; extract the trajectory motion deviation associated with the usage period from the robotic arm operation log to obtain the periodic trajectory motion deviation data; perform perception record analysis of the robotic arm motion trajectory jitter based on the periodic trajectory motion deviation data to obtain the trajectory deviation jitter perception record data; A behavior deviation feature learning module, configured to perform robotic arm trajectory behavior deviation feature learning according to the trajectory deviation jitter perception record data to obtain the trajectory behavior progressive deviation learning data; perform iterative optimization on the robotic arm trajectory tracking execution program according to the trajectory behavior progressive deviation learning data to obtain an optimized trajectory tracking execution program; A tracking control firmware design module, configured to design the robotic arm trajectory tracking control firmware for the optimized trajectory tracking execution program to obtain the robotic arm trajectory tracking control firmware; embed the robotic arm trajectory tracking control firmware into the robotic arm control center to perform robotic arm trajectory tracking control.

[0014] The beneficial effects of the present invention are as follows. By obtaining the operation logs of the robotic arm and extracting the periodic trajectory deviations, accurate analysis of the robotic arm's motion trajectory can be achieved. Through the extraction of the periodic trajectory motion deviation data, the trajectory deviations of the robotic arm in different operating cycles can be identified, and thus the regularity of its motion jitter can be perceived. This provides important basic data for subsequent analysis, enabling in-depth understanding of the subtle problems in the robotic arm's motion, helping engineers identify and diagnose potential problems in the robotic arm's motion control, and improving the motion accuracy and stability of the robotic arm. By learning and analyzing the data on the perception and recording of the trajectory deviation jitter, the progressive deviation characteristics in the robotic arm's trajectory behavior can be effectively extracted. Through the learning of the progressive deviation of the trajectory behavior, it can help the R & D team understand the accumulation trend of the trajectory errors of the robotic arm during actual operation. Based on these data, the trajectory tracking execution program of the robotic arm can be iteratively optimized, improving the trajectory tracking accuracy of the robotic arm, thereby enhancing the adaptability and accuracy of the robotic arm in complex tasks and optimizing the overall performance of the robotic arm. Through the firmware design of the optimized trajectory tracking execution program, the control ability of the robotic arm is further enhanced. During the firmware design process, various characteristics of the robotic arm's trajectory tracking are comprehensively considered to ensure that the firmware can efficiently execute control instructions and achieve precise trajectory tracking. Embedding the optimized control firmware into the robotic arm control center can adjust and precisely execute the trajectory control tasks of the robotic arm in real time, improving its execution efficiency and response speed. In addition, the optimized control firmware can also enhance the autonomous decision-making ability of the robotic arm in complex environments, improving the adaptability and operating stability of the robotic arm. Therefore, the present invention is an improvement on the traditional robotic arm trajectory tracking control method, solving the problem that the traditional robotic arm trajectory tracking control method cannot accurately learn the deviations caused by aging and wear of the robotic arm after a certain service life, resulting in large errors in the robotic arm trajectory tracking control. It improves the accuracy of learning the deviations caused by aging and wear of the robotic arm after a certain service life and reduces the errors in the robotic arm trajectory tracking control. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a schematic diagram of the step flow of a robotic arm trajectory tracking control method; Figure 2 is Figure 1 a detailed implementation step flow diagram of step S2 in Figure 3 is Figure 1 a detailed implementation step flow diagram of step S3 in DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] Please refer to Figures 1 to 3 , a robotic arm trajectory tracking control method, the method includes the following steps: Step S1: Obtain the operating log of the robotic arm and the robotic arm trajectory tracking execution program; extract the trajectory motion deviation associated with the usage cycle from the operating log of the robotic arm to obtain the periodic trajectory motion deviation data; perform the perception record analysis of the robotic arm motion trajectory jitter based on the periodic trajectory motion deviation data to obtain the trajectory deviation jitter perception record data; Step S2: Perform the learning of the robotic arm trajectory behavior deviation characteristics according to the trajectory deviation jitter perception record data to obtain the trajectory behavior progressive deviation learning data; perform the iterative optimization of the trajectory tracking control on the robotic arm trajectory tracking execution program according to the trajectory behavior progressive deviation learning data to obtain the trajectory tracking execution optimization program; Step S3: Design the robotic arm trajectory tracking control firmware for the trajectory tracking execution optimization program to obtain the robotic arm trajectory tracking control firmware; embed the robotic arm trajectory tracking control firmware into the robotic arm control center to execute the robotic arm trajectory tracking control.

[0017] In the embodiment of the present invention, refer to Figure 1 As described, it is a schematic diagram of the step flow of a robotic arm trajectory tracking control method of the present invention. In this example, the robotic arm trajectory tracking control method includes the following steps: Step S1: Obtain the operating log of the robotic arm and the robotic arm trajectory tracking execution program; extract the trajectory motion deviation associated with the usage cycle from the operating log of the robotic arm to obtain the periodic trajectory motion deviation data; perform the perception record analysis of the robotic arm motion trajectory jitter based on the periodic trajectory motion deviation data to obtain the trajectory deviation jitter perception record data; In the embodiments of the present invention, by configuring the PLC acquisition system and the embedded industrial gateway in the industrial field, the operation log data of the robotic arm under different working conditions is collected and stored in the local database in the form of millisecond-level timestamp synchronization. The operation log includes high-frequency control quantities such as the angle changes of the six joints of the robotic arm, the spatial pose of the end effector, speed, acceleration, joint torque, joint current, feedback control signal, and instruction execution feedback. The synchronized robotic arm trajectory tracking execution program is modularly parsed in the form of structured code blocks to extract the corresponding motion control instruction set and trajectory interpolation function. Next, in combination with the time period information in the robotic arm operation log, the operation log data is segmented based on the period boundary points and segmented according to the task execution period set in the process flow. The differences between the actual trajectory data and the instruction trajectory data within each period segment are sampled and calculated, the trajectory deviation metric is calculated through the vector Euclidean distance, and statistical analysis is performed after normalization at each key point to form the periodic trajectory motion deviation data. Then, based on the periodic deviation data, through the sliding window time series feature analysis technology, the change trend of the position deviation of the trajectory in consecutive periods is extracted, and a trajectory jitter amplitude change trajectory is established using a combination model of moving average and weighted difference. After establishing a three-dimensional jitter data matrix (position deviation, speed change, acceleration mutation) in the time series dimension, principal component analysis (PCA) and Mahalanobis distance analysis are performed to calibrate the outlier period points of the trajectory deviation and their trend clustering intervals, and finally, trajectory deviation jitter perception record data is generated.

[0018] Step S2: Perform robotic arm trajectory behavior deviation feature learning based on the trajectory deviation jitter perception record data to obtain trajectory behavior progressive deviation learning data; perform trajectory tracking control iterative optimization on the robotic arm trajectory tracking execution program according to the trajectory behavior progressive deviation learning data to obtain a trajectory tracking execution optimization program; In the embodiments of the present invention, trajectory behavior deviation feature learning processing is performed on the trajectory deviation jitter perception record data. The specific method is as follows: perform time-series frequency spectrum transformation on the jitter data matrix of each period, and use fast Fourier transform (FFT) and wavelet packet decomposition (WPD) to extract the main frequency components of trajectory deviation and the change trend of jitter amplitude in different frequency bands. In the spectrum analysis, set the frequency resolution to 0.5 Hz, and record the change in the proportion of the energy of each spectrum segment. Next, establish a progressive deviation trend sequence of trajectory behavior, and judge whether there is a periodic deviation accumulation trend by calculating the cosine similarity of the jitter intensity vectors between periods. Map the trend data to the progressive deviation learning dataset of trajectory behavior, and use the clustering division method (such as K-Means) to classify the periods with similar deviation behaviors, label their offset vector fields and historical trend fields, and establish trajectory behavior deviation data labels. Subsequently, perform reconstruction processing on the original trajectory tracking execution program of the robotic arm. According to the trajectory key point deviation vector field marked in the learning data, reset the angular offset of the trajectory execution point by introducing an interpolation correction control node method, and perform spline reconstruction processing in combination with the trajectory planning module. Add a feedback suppression link to the reconstructed trajectory program, introduce the real-time correction coefficient for the deviation estimation amount in the PD controller, and perform local gain optimization to finally obtain the optimized trajectory tracking execution program.

[0019] Step S3: Design the robotic arm trajectory tracking control firmware for the optimized trajectory tracking execution program to obtain the robotic arm trajectory tracking control firmware; embed the robotic arm trajectory tracking control firmware into the robotic arm control center to perform robotic arm trajectory tracking control.

[0020] In the embodiments of the present invention, a static code analysis tool (such as the Clang static analyzer) is used to perform code-level control logic path analysis on the trajectory tracking execution optimization program, identify the dependency chains and jump logics between execution nodes in all path branches, and detect whether there are potential blocking paths or loop abnormal calls. After the analysis, the integrity of the execution logic is verified by establishing a control flow graph (CFG). Subsequently, based on the static analysis results, a set of functional unit test cases is constructed to perform module-level integration testing on the program. The test contents include: the point output accuracy test of the trajectory reconstruction module (setting the maximum allowable error to 0.05 mm), the real-time responsiveness of the feedback correction module (testing that the deviation correction delay does not exceed 0.5 ms under a 1 ms control cycle), the robustness of the control parameter real-time scheduling module, etc. The test uses a Python automation script to link with the PLC simulation system to run and record the output results. After all test data meet the system robustness constraints, the optimization program is packaged as a trajectory control firmware. This firmware is compiled in the C language embedded executable format and is adaptively configured according to the register mapping and interrupt vector table of the target controller (such as the STM32H7 series chips based on the ARM Cortex-M7 kernel). The control firmware is written into the main control unit of the robotic arm through the JTAG interface. When running in the control center, the trajectory control firmware will be automatically loaded and executed in each cycle control scheduling, replacing the original control program to complete the trajectory tracking control task with higher accuracy and faster response.

[0021] Step S1 includes the following steps: Step S11: Obtain the operation log of the robotic arm and the execution code of the robotic arm trajectory tracking; Step S12: Clean the data of the operation log of the robotic arm to obtain the cleaned operation log of the robotic arm; Step S13: Extract the trajectory motion deviation associated with the usage cycle from the cleaned operation log of the robotic arm to obtain the periodic trajectory motion deviation data; Step S14: Based on the periodic trajectory motion deviation data, perform analysis on the perception record of the robotic arm motion trajectory jitter of the cleaned operation log of the robotic arm to obtain the trajectory deviation jitter perception record data.

[0022] In the embodiments of the present invention, the process of obtaining the operating log of the robotic arm and the trajectory tracking execution code is to configure the industrial Ethernet communication interface, connect to the internal log system of the robotic arm controller, and transmit the operating log data to the local data acquisition end in the form of TCP / IP protocol. The log content includes high-frequency data such as the six-axis joint angles, speeds, accelerations, torques, currents, Cartesian coordinate poses of the end effector, system timestamps, target trajectory point sequences, and feedback errors output by the controller. The sampling period is 10 ms, and the data accuracy is four decimal places. The trajectory tracking execution code is written in PLC structured text language, including components such as trajectory interpolation functions, joint space-Cartesian space conversion functions, PID controller configuration parameters, trajectory point planning modules, and fault tolerance processing logics. After the code is exported from the controller storage area through a dedicated interface, it is parsed at the module level, and the key instruction sequences are aligned with the log data by timestamps to provide a data basis for subsequent steps. Clean the obtained operating log data. First, use an outlier detection algorithm to remove outliers from each data column in the log. The method of detecting outliers based on the median absolute deviation (MAD) is adopted. Set the threshold to 1.5 times the IQR interval, and remove the data points outside this interval in each column. Then, unify the timestamp format, resample all data to a fixed interval of 10 ms, and use linear interpolation to fill in the missing data. Remove duplicate record entries during the sampling process, and select unique entries with the system clock field in the log as the primary key. And perform unit conversion for all numerical fields, for example, unify the angle to radian system and the displacement to millimeter system. After cleaning, construct a standard structured operating cleaning log containing fields such as timestamps, six-axis joint states, end effector positions and postures, and controller output signals. The output format is a CSV file, and the field order is kept consistent with the control logic. Extract the trajectory motion deviation associated with the usage cycle based on the operating cleaning log of the robotic arm. First, mark the cycle boundaries of the operating log through the timestamps corresponding to the start instruction and stop instruction recorded in the trajectory tracking execution code. Extract the target trajectory points and actual feedback trajectory points in each usage cycle, and calculate the three-dimensional Euclidean distance in the Cartesian space for each trajectory point as the instantaneous trajectory error value according to the timestamps. Normalize the error value vector of each cycle, and apply a first-order difference operation to the error curve to analyze its change trend. Adopt a periodic sliding window mechanism to smooth the trajectory errors within three consecutive cycles by weighted mean, and remove the high-frequency disturbance terms, and retain the trend deviation component. Finally, count the maximum deviation value, average deviation value, and trajectory deviation change slope of each cycle, construct a periodic trajectory motion deviation data set, and output it in a structured form as the input basis for subsequent trajectory jitter perception analysis. Combine the periodic trajectory motion deviation data with the operating cleaning log to complete the analysis of the robotic arm motion trajectory jitter perception record.The analysis method is based on the frequency-domain analysis technology of time-series signals. First, perform a fast Fourier transform (FFT) on the trajectory deviation time series of each cycle, extract the main frequency peak frequency and its amplitude, and count the energy proportion of the top three frequency components. Then, introduce wavelet packet decomposition (WPD) to perform multi-scale decomposition on the trajectory deviation sequence, extract the energy fluctuations in the frequency bands of 2 Hz, 4 Hz, and 8 Hz, and use them as important indicators of the change in the trajectory oscillation frequency. Construct a three-dimensional jitter perception vector (main frequency, secondary frequency, frequency mutation rate) based on the change in frequency components, and compare it with the acceleration change curve in the corresponding cycle to identify the trajectory perturbation events corresponding to the synchronous frequency mutation. By setting the frequency mutation rate threshold to 20% and the jitter peak change rate threshold to 15%, screen out the cycle segments with significant trajectory jitter and mark their cycle numbers and jitter types (continuous type, burst type, low-frequency type). Finally, construct a trajectory deviation jitter perception record data table containing time period, main frequency component, amplitude change, mutation rate, and perturbation type, and store it in JSON format to provide stable basic data for subsequent deviation feature learning.

[0023] Step S2 includes the following steps: Step S21: Analyze the skewness intensity of the trajectory jitter amplitude change for the trajectory deviation jitter perception record data to obtain the trajectory jitter amplitude skewness intensity data; Step S22: Perform progressive analysis of the spatial deviation amount on the periodic trajectory motion deviation data to obtain the progressive data of the deviation amount between trajectory motions; Step S23: Perform progressive deviation feature learning of the robotic arm trajectory behavior based on the trajectory jitter amplitude skewness intensity data and the progressive data of the deviation amount between trajectory motions to obtain the progressive deviation learning data of the trajectory behavior; Step S24: Perform iterative optimization of the trajectory tracking control on the robotic arm trajectory tracking execution program based on the progressive deviation learning data of the trajectory behavior to obtain the optimized trajectory tracking execution program.

[0024] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: Analyze the skewness intensity of the trajectory jitter amplitude change for the trajectory deviation jitter perception record data to obtain the trajectory jitter amplitude skewness intensity data; In the embodiments of the present invention, the trajectory jitter amplitude change skewness intensity of the trajectory deviation jitter perception record data is analyzed, and the processing is completed by combining the statistical analysis and the calculation of the sliding window amplitude offset. In the specific operation process, first, the amplitude corresponding to the main frequency of the trajectory deviation signal in each period is extracted cycle by cycle, and the amplitude value of the frequency domain peak point is used as an index to form a trajectory amplitude time series. The sequence is divided into analysis windows in units of 10 cycles, and indicators such as skewness, kurtosis, ratio of maximum value to mean value, and ratio of standard deviation to mean value of the amplitude are calculated in each window. Skewness is used to characterize the offset degree of the amplitude distribution on both sides of the central value, kurtosis is used to describe the frequency of abnormal sudden increase in the amplitude value, and the ratio of standard deviation to mean value reflects the stability fluctuation range. For the window data in different time periods, by setting the combined judgment condition that the absolute value of skewness is greater than 1.5 and kurtosis is greater than 3.5, the trajectory jitter behavior in this window is marked as having a skewness intensity feature. Record the starting cycle number, average amplitude value, maximum offset amplitude, amplitude growth rate, and fluctuation frequency of all windows with skewness features as the main fields of the trajectory jitter amplitude skewness intensity data. This data structure is uniformly stored as a structured two-dimensional array, with the field precision controlled to three decimal places and the time synchronization precision not less than 10 ms.

[0025] Step S22: Perform a progressive analysis of the spatial deviation amount on the periodic trajectory motion deviation data to obtain the progressive data of the deviation amount between trajectory motions; In the embodiments of the present invention, the progressive analysis of the spatial deviation amount on the periodic trajectory motion deviation data is completed by establishing a relative deviation increment model between trajectory points. The specific method is to extract N fixed trajectory points (for example, sampling points at every 10 mm interval of the end effector) in each period, record their target position and actual position coordinates respectively, and calculate the three-dimensional spatial Euclidean deviation value for each sampling point. Perform a difference operation on the deviation value sequence of the same points in consecutive periods to obtain a deviation increment vector. Use the linear fitting method to fit the trend of the deviation increment of each point evolving with the period, and extract the slope, standard deviation of the deviation increase, and change rate of the increment direction as progressive indicators. For points with a slope greater than a certain threshold (such as 0.005 mm / cycle) and a standard deviation less than the threshold (such as 0.1 mm), it is judged that their spatial deviation has an asymptotic accumulation trend. Finally, extract the statistical indicators of all progressive points in the entire period segment, such as the proportion of progressive points, the maximum progressive increment, and the progressive direction consistency coefficient (the direction vectors have an included angle less than 15 degrees to be consistent), to form the progressive data of the deviation amount between trajectory motions. The data structure includes fields such as cycle number, trajectory segment number, number of progressive points, and progressive parameters of each point, and is sorted and stored according to the cycle time axis.

[0026] Step S23: Perform progressive deviation feature learning on the robotic arm trajectory behavior based on the trajectory jitter amplitude skewness intensity data and the progressive data of the deviation amount between trajectory movements, to obtain the progressive deviation learning data of the trajectory behavior; In the embodiment of the present invention, progressive deviation feature learning on the robotic arm trajectory behavior is performed based on the trajectory jitter amplitude skewness intensity data and the progressive data of the deviation amount between trajectory movements, and features are extracted through multivariate trajectory behavior vector encoding and principal component clustering analysis method. First, for each period, a trajectory behavior vector including five dimensions of amplitude main frequency, amplitude skewness intensity, progressive slope, progressive direction consistency, maximum progressive increment, etc. is constructed, and all vectors form a behavior data matrix. After scaling all field values to the interval [0,1] by using standardization processing, the covariance matrix is calculated, and based on this matrix, principal component analysis is performed to extract the first two principal components with a cumulative contribution rate exceeding 95%. The dimension-reduced behavior vectors are input into the density-based clustering method (DBSCAN) for unsupervised clustering analysis, with the minimum number of samples set to 5 and the density threshold set to 0.3, to identify different categories of behavior deviation patterns. For each category, the typical period segment number is marked, the representative behavior parameter values are recorded, and the progressive deviation learning data of the trajectory behavior is generated, including fields such as principal component distribution, behavior cluster identification, representative period, mean and variance of behavior indicators, etc. This data is output as a CSV format file, which is used to guide the optimization direction of the control program.

[0027] Step S24: Perform iterative optimization of the trajectory tracking control on the robotic arm trajectory tracking execution program according to the progressive deviation learning data of the trajectory behavior, to obtain the optimized trajectory tracking execution program.

[0028] In the embodiments of the present invention, the trajectory tracking control iteration optimization of the robotic arm trajectory tracking execution program is carried out according to the trajectory behavior progressive deviation learning data, and is realized by combining controller parameter adjustment and trajectory interpolation strategy adjustment. First, analyze the controller PID parameter configuration of the cycle segment corresponding to each type of deviation behavior, and record the historical setting values of the three parameters P, I, and D and the corresponding deviation performance. For the cycle segment with progressive deviation, increase the value of the P parameter to enhance the controller response ability, and at the same time reduce the value of the I parameter to suppress the integral accumulation lag, and set the adjustment step size to 5% of the initial value. For the cycle segment with high-frequency jitter characteristics, reduce the value of the D parameter to suppress the oscillation influence, and at the same time optimize the trajectory point spacing, adjust the interpolation point distribution to a denser style, so that the end pose change is smoother. After all parameter adjustments, overwrite and update the trajectory interpolation function segment and the PID controller configuration segment in the control program code, and retain the adjusted historical version for the basis of comparative experiments. Finally, execute and verify the modified code in the simulation environment, record the changes in the trajectory error and jitter index of each cycle after optimization, confirm the performance improvement, and then solidify the code version to form the trajectory tracking execution optimization program. All control parameters in this program are encapsulated in the form of a structure, recording the version number and the cycle number of the parameter source for traceability analysis and subsequent iteration.

[0029] Step S21 includes the following steps: Step S211: Perform time-frequency domain conversion on the trajectory deviation jitter perception record data to generate a deviation jitter time-frequency domain diagram; Step S212: Perform skewness recognition of the distribution of jitter frequency mutation points on the deviation jitter time-frequency domain diagram to obtain skewness data of the distribution of jitter frequency mutation points; Step S213: Calculate the variance of the increasing slope of the mutation according to the skewness data of the distribution of jitter frequency mutation points to obtain the variance of the increasing slope of the frequency mutation points; Step S214: Perform convergence constraint logarithmic fitting processing on the deviation jitter time-frequency domain diagram based on the variance of the increasing slope of the frequency mutation points to obtain convergence logarithmic fitting data of the frequency mutation points; Step S215: Analyze the skewness intensity of the trajectory jitter amplitude change according to the convergence logarithmic fitting data of the frequency mutation points to obtain skewness intensity data of the trajectory jitter amplitude.

[0030] In the embodiments of the present invention, time-frequency domain conversion is performed on the trajectory deviation jitter perception record data, and the discrete Fourier transform (DFT) method is used to perform frequency domain mapping processing on the mechanical arm end trajectory deviation displacement sequence within each period. Specifically, during processing, the trajectory deviation data collected within each period is composed into a real number sequence with equal time intervals in chronological order, 1024 sampling points per period are intercepted at a sampling frequency of 1000 Hz, and the fast Fourier transform (FFT) algorithm is applied to this time series to obtain the corresponding frequency amplitude spectrum diagram, with the frequency range being from 0 Hz to 500 Hz. The frequency domain diagram uses frequency as the horizontal axis and amplitude as the vertical axis, and the storage format is a two-dimensional array. Each element in the array contains the frequency value and the corresponding amplitude value, with the precision set to three decimal places. The time-frequency domain diagram is numbered according to the period number and archived in a unified directory structure. Skewness recognition of the jitter frequency mutation point distribution is performed on the deviation jitter time-frequency domain diagram, using a combination method of first-order difference mutation point detection and peak counting. Specifically, first-order difference processing is performed on the frequency axis of the frequency domain diagram for each period, and the frequency positions where the difference amplitude is greater than a set threshold (such as 0.15) are identified as the mutation point positions. Subsequently, local peak judgment is performed on the 3 frequency components before and after each mutation point, with the judgment criterion being that the current amplitude is greater than the amplitudes before and after and more than twice the average amplitude of the entire spectrum. The frequency distribution of the mutation point positions in each period is statistically analyzed, a frequency histogram of the mutation frequency positions is plotted, and its skewness and kurtosis are calculated. A skewness greater than 1 indicates that the distribution deviates from low-frequency concentration, and a kurtosis greater than 4 indicates that the mutation frequencies are concentrated in a certain frequency band. The skewness and kurtosis are combined into a structure, which serves as the skewness data of the jitter frequency mutation point distribution for each period. This structure contains a list of mutation point frequency values, an average amplitude value, a frequency skewness value, and a frequency kurtosis value.

[0031] Calculate the variance of the mutation increasing slope based on the skewness data of the mutation frequency mutation point distribution, and process it using the mathematical statistics method of combining slope fitting and variance analysis. Perform linear fitting on the amplitude evolution trend of the same mutation point frequency in multiple consecutive periods to obtain the amplitude increasing slope sequence of each mutation point frequency. Calculate the standard deviation of the slope sequence to obtain the increasing slope fluctuation intensity of the frequency mutation point. To improve the analysis accuracy, divide the mutation frequency range into small intervals of 10 Hz each, and calculate the slope mean and variance of the mutation points within each group respectively. If the variance value is greater than the preset threshold (such as 0.05), it is determined that the mutation increasing fluctuation in this frequency band is strong. The slope variance values corresponding to all mutation frequency bands form the data set of the increasing slope variance of the frequency mutation point, and the fields include the start and end values of the frequency band, slope variance, number of mutation frequency points, and average increasing amplitude, etc. Perform convergence constraint logarithmic fitting processing on the deviation jitter time series frequency domain diagram based on the increasing slope variance of the frequency mutation point, and adopt the logarithmic fitting curve and residual convergence limitation strategy. Select the frequency domain diagram data of each period within the mutation frequency band with a higher variance value, extract the data sequence of the mutation point amplitude evolving with the period, and construct the fitting data pair between the period number and the amplitude. Perform fitting in the form of natural logarithm, use the least squares method to solve the parameters of the logarithmic function, and constrain the fitting residual not to exceed 0.1 to control the convergence error. Determine that the absolute value of the derivative of the fitting curve is continuously less than 0.01 in the last five periods as the convergence condition, and record the frequency band and corresponding fitting parameters that meet this condition as the convergence logarithmic fitting data of the frequency mutation point. The data structure includes the frequency band range, fitting constant parameters, maximum residual, and fitting convergence period number, etc. Analyze the skewness intensity of the trajectory jitter amplitude change based on the convergence logarithmic fitting data of the frequency mutation point, and adopt the frequency weight integration and high amplitude point ratio extraction method. For each frequency band within the fitting convergence period, extract the amplitude growth rate corresponding to this frequency band and perform normalization processing as the amplitude change speed factor. At the same time, count the number of points with an amplitude greater than twice the overall average value in each frequency band, and calculate the high amplitude point ratio with the total number of points as the denominator. Finally, combine the amplitude growth rate factor and the high amplitude point ratio, and calculate the skewness intensity index of this frequency band through the weighted integration method. After merging the skewness intensity index data of all frequency bands, index them according to the frequency band number and the corresponding period number to form the data set of the skewness intensity of the trajectory jitter amplitude. The data accuracy is reserved to three decimal places, and the fields include the frequency band range, skewness intensity value, maximum amplitude value, high amplitude ratio, fitting residual, etc.

[0032] Step S214 includes the following steps: Perform mutation point frequency domain curve increasing segmented exponential transformation processing on the deviation jitter time series frequency domain diagram based on the increasing slope variance of the frequency mutation point to obtain the mutation point curve increasing segmented exponent; Perform increasing segmented exponential ratio convergence processing on the mutation point curve increasing segmented exponent to obtain the increasing segmented exponential convergence ratio; Perform a convergence constraint logarithmic fitting process on the deviation jitter time-frequency domain diagram according to the increasing segmented exponential convergence ratio to obtain the convergence logarithmic fitting data of the frequency mutation point.

[0033] In the embodiments of the present invention, based on the variance of the increasing slope of the frequency mutation points, a mutation point frequency-domain curve increasing piecewise exponential transformation is performed on the deviation jitter time-series frequency-domain diagram, and a method of performing piecewise exponential fitting on the corresponding frequency-domain amplitude sequence of each period within the mutation point frequency segment is used to achieve a non-linear enhanced expression of the frequency-domain fluctuation change. In specific operations, first, the frequency interval where the frequency mutation point is located is determined. For example, if the mutation point frequency is 135 Hz, an interval [125 Hz, 145 Hz] formed by ±10 Hz above and below it is used as the analysis range. The amplitude sequences in this frequency interval for N = 10 consecutive periods are divided, with every 3 periods as a segment, forming three data segments, and each segment contains the amplitude sequences corresponding to the same frequency points. Each segment is fitted using the form of a power exponential fitting function, and the least squares method is used to determine the fitting parameters, where the change rate of the exponential part is used as the increasing exponential value of this segment. The increasing exponential values after fitting for each segment are respectively recorded, and their corresponding period ranges and frequency ranges are marked to form a mutation point curve increasing piecewise exponential data structure, which includes fields: start and end values of frequency, start and end values of period, fitting exponential parameter, sum of squared fitting residuals, position of the maximum frequency-domain amplitude point, etc. An increasing piecewise exponential ratio convergence process is performed on the mutation point curve increasing piecewise exponential, and a method of decreasing determination of the piecewise exponential ratio and analysis of the change rate within the convergence window is used to achieve the stability detection of the frequency-domain fluctuation intensity. The specific method is to calculate the ratio of the increasing exponential values of two adjacent segments. For example, if the exponential value of the previous segment is α1 and the exponential value of the next segment is α2, calculate α2 / α1 as the ratio of this comparison segment. If the ratio is less than 1 in two consecutive segments and the change amplitude is less than the set threshold (such as 0.05), it is determined that a convergence trend appears. At the same time, the absolute value of the ratio change rate is calculated within a three-segment sliding window. If the maximum rate does not exceed 0.03, it is considered that the change of the increasing exponential value is in a stable state. The frequency band and period range that meet this condition are marked as the increasing exponential convergence region, and increasing piecewise exponential convergence ratio data is output, which includes fields: in-segment exponential value, inter-segment ratio, convergence period window number, ratio stable rate, residual ratio change amplitude, etc. According to the increasing piecewise exponential convergence ratio, a convergence constraint logarithmic fitting process is performed on the deviation jitter time-series frequency-domain diagram, and a convergence-weight adjusted logarithmic fitting algorithm is used to improve the accuracy of the convergence model for the energy change of the frequency mutation points.In a specific implementation, for the frequency band data of all the identified exponentially decaying regions, the amplitude change sequence of the same frequency point during the periodic evolution is extracted, and a fitting data set with the cycle number as the independent variable and the amplitude as the dependent variable is constructed. A natural logarithm function form is used for fitting, and the function form is set as y = a×ln(t) + b, where t is the cycle number, that is, the number of cycle executions of the corresponding trajectory; y represents the frequency domain amplitude response intensity at a specific frequency point, that is, the change of y with t represents the evolution trend of the robotic arm trajectory jitter amplitude at this frequency under consecutive cycles; a represents the growth or decay rate of the amplitude with the change of the cycle number; b represents the reference amplitude offset (initial logarithmic amplitude) at t = 1. To ensure the fitting convergence, a higher fitting weight is given to the segment with a smaller rate of change of the ratio. The weighted least squares fitting method is used to determine the logarithmic function parameters. After the fitting is completed, the residuals of all the fitting curves are statistically analyzed. If the mean square value of the residuals is less than 0.08 and the amplitude change amount in the last three cycles is less than 0.01, the fitting result is recorded as a valid converging fitting curve. Finally, the converging logarithmic fitting data set of the frequency mutation points is output, and the data structure includes fields: frequency range, start and end cycle numbers, fitting parameters, weight coefficient, mean square of residuals, number of cycles in the fitting stable window, etc. This data will be used in the trajectory behavior deviation modeling stage to extract the characteristics of the stable change of the trajectory control deviation with the cycle.

[0034] In another embodiment, the variance data of the increasing slope of the frequency mutation points is divided into 24 time windows in chronological order, and the length of each window is 5 seconds. The frequency domain data within each time window is subjected to an exponential transformation with a transformation base of 2.718 to map the frequency values into the exponential space. Mutation points are extracted in the exponential space, and the mutation point determination threshold is set to 1.5 times the local mean. The exponential mean of the mutation points is calculated for each time window to form 24 exponential mean sequences. The exponential mean sequences are segmented, and the segmentation points are set at positions where the change rate of the exponential value exceeds 30%. Linear regression is performed on each segment of data to obtain the segment slope. The segment slopes, exponential means, and the number of mutation points of the 24 time windows are combined into a feature vector to form the increasing segment exponential of the mutation point curve. The increasing segment exponential of the mutation point curve is normalized to unify the numerical range to the 0-1 interval. The ratio sequence of the exponential values between adjacent segments is calculated. The ratio convergence threshold is set to 0.85, and it is determined to converge when the change rate of three consecutive ratios is less than 15%. The non-converged segments are iteratively processed, and the segment length is halved each time, and the exponential ratio is recalculated. The upper limit of the number of iterations is set to 8 times. The ratio sequence and convergence status of each iteration are recorded. The finally converged ratio sequence, convergence position, and number of iterations are integrated into feature data to obtain the converged ratio of the increasing segment exponential. Based on the converged ratio of the increasing segment exponential, logarithmic fitting constraint conditions are constructed, and the constraint conditions include a maximum fitting error of 0.1, a minimum convergence ratio of 0.8, and a maximum number of iterations of 12 times. The original frequency domain map data is segmented according to 5-second time windows. The baseline drift is subtracted from each segment of data, and the baseline is obtained through 60-point sliding median filtering. Logarithmic function fitting is performed on the baseline-removed data, and the trust region method is used to solve for the optimal fitting parameters. The convergence ratio is introduced as a weight during the fitting process, and the larger the convergence ratio, the higher the weight. The fitting stops when the fitting error is less than the threshold or the maximum number of iterations is reached. The fitting results of all segments are stitched together along the time dimension to obtain the converged logarithmic fitting data of the frequency mutation points.

[0035] Step S23 includes the following steps: Step S231: Calculate the progressive mean difference of the attitude angle offset for the progressive data of the deviation amount between trajectory movements to obtain the progressive mean difference of the attitude angle offset; Step S232: Calculate the incremental ratio of the joint angle deviation time series for the progressive data of the deviation amount between trajectory movements to obtain the incremental ratio of the joint angle deviation time series; Step S233: Perform a robotic arm trajectory jitter swing regression analysis on the progressive mean difference of the attitude angle offset and the incremental ratio of the joint angle deviation time series according to the trajectory jitter amplitude skewness intensity data to obtain the robotic arm trajectory jitter swing regression data; Step S234: Perform transfer learning on the robotic arm trajectory jitter swing regression data for repeated trajectory jitter swings to obtain the repeated trajectory jitter swing learning data; Step S235: Perform progressive deviation feature learning on the robotic arm trajectory behavior based on the repeated trajectory jitter swing learning data to obtain trajectory behavior progressive deviation learning data.

[0036] In an embodiment of the present invention, the progressive mean difference of the attitude angle offset is calculated for the progressive data of the deviation between trajectory motions, and the change characteristics of the attitude angle of the end effector in each cycle are obtained by statistical methods. The specific process is: extract the attitude angle (Euler angle around the X, Y, and Z axes) offset value sequence of each sampling point in 20 consecutive cycles from the deviation data set between trajectory motions, and apply sliding window analysis to the three angle directions of X, Y, and Z respectively, with the window length set to 5 cycles and the step length set to 1 cycle. Calculate the offset mean for each angle direction in the window, and extract the deviation mean difference, that is, the mean difference between the absolute value of the attitude angle offset in the window and the window mean. The mean difference value of each window is summarized in the entire time period, and its mean and standard deviation are counted. The result is the progressive mean difference of the attitude angle offset, in degrees, with accuracy controlled to three decimal places, and record the corresponding trajectory segment number and cycle number, and output in CSV format. The joint angle deviation time series increment ratio is calculated for the progressive data of the deviation between trajectory motions, and the cumulative trend of the angle error of each joint is revealed through normalization processing. Specific method: Extract the continuous 20-cycle deviation value sequence of each sampling endpoint of the six joints from the data set, calculate the time series increment for each joint, that is, the difference of adjacent cycle deviations, and obtain 19 increment value sequences. For each joint, the increment sequence is positive according to the absolute value, and the percentage ratio between each increment value and the maximum increment value of the joint sequence is calculated. Then the average increment ratio and variance of each joint are calculated to form the joint angle deviation time series increment ratio data. The data format includes: cycle number, joint number, average increment ratio, increment variance, maximum increment cycle number and other fields, with accuracy to three decimal places. According to the trajectory jitter amplitude skewness intensity data, the progressive mean difference of the attitude angle offset and the time series increment ratio of the joint angle deviation are used for trajectory jamming amplitude regression analysis to identify the jamming swing behavior characteristics. The jamming amplitude prediction model is established by the multivariate linear regression method, and the independent variables include amplitude skewness intensity, progressive mean difference of the attitude angle and the average increment ratio of each joint. The dependent variable is the end effector posture jitter amplitude (unit: mm). The data set comes from multiple cycle segments, and 200 sets of data are selected for training from the 7th to the 27th cycle. The least squares method is used to estimate the regression coefficient, and the model R2R^2R2 value and parameter significance p value are calculated. The regression coefficient corresponding to each independent variable must satisfy the p value <0.05 and R^2 greater than 0.85; the regression process removes outliers with residuals exceeding twice the standard deviation. The regression result is output as the regression data of the robot trajectory jamming swing, and the format includes regression coefficient, p value, R^2, residual standard deviation, and sampling period range. Repeated trajectory jamming swing regression data is used to perform trajectory jamming swing transfer learning, extract the similarity of the jamming swing behavior of the previous running trajectories, and realize the extraction of behavioral migration features. The specific method is: the jamming swing regression data collected from multiple trajectory runs are classified according to the trajectory type, such as multiple runs of the same processing path.For each type of trajectory operation data set, use dynamic time warping (DTW) to align and compare the time series of regression coefficients for each operation, calculate the DTW distance between each pair, and determine that the repeated trajectory behaviors are similar if the distance is less than a threshold (such as 0.2). Count the processes in each type of trajectory where the number of similarities exceeds 3 times, extract the average coefficient and standard deviation of the corresponding regression model, and form the learning data of the repeated trajectory stuttering swing amplitude, with fields including trajectory ID, average regression coefficient, coefficient variance, number of similarities, and mean DTW distance. Conduct progressive deviation feature learning of the robotic arm trajectory behavior based on the learning data of the repeated trajectory stuttering swing amplitude, and map the identified repeated stuttering swing features to the potential deviation patterns of the trajectory behavior. Use the clustering analysis method to perform unsupervised clustering on the learning data of the repeated trajectory stuttering swing amplitude corresponding to different trajectory IDs, adopt the Ward hierarchical clustering method, calculate the Euclidean distance between each data point, and construct a clustering tree. Set the number of clusters for the clustering result to be between 3 and 5, and determine the optimal number of clusters in combination with the between-class sum of squares error analysis (Elbow method). Calculate the center point for each cluster, which is the progressive deviation feature vector of the trajectory behavior of this class, and assign a cluster label. Finally, generate the progressive deviation learning data set of the trajectory behavior, including fields such as trajectory cluster ID, central regression coefficient, within-cluster variance, cluster member period range, similarity score, etc. The data format is in JSON structure and is used as the basis for the subsequent trajectory tracking control optimization program design.

[0037] In another embodiment, the periodic trajectory motion deviation data is first called. In a complete task cycle, frame sequence sampling is performed according to the angular errors of the six joints of the robotic arm. The sampling frequency is set to 1000 Hz, the trajectory segment length is 5 seconds, and the total number of trajectory broken lines sampled is 5000 frames. Euler angle conversion processing is performed on the difference between the desired trajectory and the actual trajectory of the end effector pose of the six axes during adjacent trajectory execution cycles to calculate the corresponding attitude angle error vector. Then, by calculating the average growth of the attitude difference vector in the three attitude angle axes (pitch, yaw, roll) directions in each adjacent execution cycle, a data block with a sliding window length of 10 cycles is successively used to calculate its mean difference change, and the autoregressive moving average filtering method is used to smooth the change trend. On all data blocks, its progressive mean difference is calculated and the change index is recorded. Finally, an attitude angle offset progressive mean difference sequence is output, in the format of a three-dimensional array. Each axis includes statistical quantities such as frame number, continuous cycle mean difference, mean increase rate, local extreme position index, etc. Temporal increment processing of joint angle deviation is performed on the progressive data of the deviation amount between trajectory motions corresponding to step S231. Specifically, the angular error value of each joint of the robotic arm is used as a time series, and a first-order difference is taken between adjacent frames to obtain an angle increment sequence. After dividing the increment by the theoretical angular change of the previous frame, the increment ratio is obtained. Subsequently, a time sliding window process is performed on each frame. The window is set to 100 frames, that is, corresponding to 100 ms. The average value and variance of the ratio of the joint angle deviation increment to the theoretical speed in this segment are statistically calculated, and the part exceeding the set threshold (such as ±15%) is screened and marked as a sensitive site. Finally, a temporal increment ratio sequence of joint angle deviation is generated, including data fields: joint number, frame number, deviation increment ratio, local mutation point position, mutation index, etc., for judging the response inconsistency rate between the control instruction and the feedback execution. Based on the trajectory jitter amplitude skewness intensity data as an adjustment factor, the attitude angle offset progressive mean difference and the temporal increment ratio data of joint angle deviation are jointly input to construct a regression analysis system for robotic arm trajectory jitter and swing amplitude. The regression analysis method uses a multiple linear regression enhancement model. The main process is to use the identified abnormal jitter intensity segments in each cycle as target labels, and the attitude mean difference and offset ratio in the corresponding cycle as independent variables. By setting an attribution threshold of 0.75, it is judged whether the correlation joint degree of the results meets the fitting conditions. At the same time, the Bayesian information criterion (BIC) is embedded in the regression model to perform conditional screening on each dependent variable to eliminate non-significant variables. Finally, the target fields are output: regression residual, list of fitting linear equations, variable weight degree ranking, regression goodness of fit, abnormal discrimination rate, cycle span mapping index, to describe the relationship between jitter and swing amplitude of the robotic arm due to structural disturbance and control lag in a specific frequency domain. Based on the trajectory jitter and swing amplitude regression data generated in multiple historical cycles, a repeated trajectory jitter and swing amplitude transfer learning process is performed.This step uses statistical learning methods to construct typical feature templates. Through the jamming data in the long-term running trajectory, the time series distribution of key change points, the axial contribution weights and the pattern of structural changes between periodicities are extracted to establish the trajectory behavior feature migration mapping structure. In the specific implementation, the jamming amplitude features verified in multiple historical trajectory files are projected into the unified time domain for normalized expression, and several typical attitude offset templates and angle response misalignment feature clusters are determined by the K-means clustering algorithm. The total number of clusters is generally set between 7 and 10, and the optimal configuration is selected based on the Davies–Bouldin index. After iterative convergence, the potential trajectory period mapping relationship representing the repeated occurrence position of jamming is accurately identified, and the jamming amplitude learning data of repeated trajectories is output, including the feature cluster mean vector, cluster center ID, feature sequence period position sequence number and corresponding error contribution index. The jamming amplitude learning data of repeated trajectories is input into the feature learning module, and the trajectory behavior progressive deviation learning data set is constructed based on the typical periodic trajectory data. Specifically, the sequences showing repeated jamming fragments are sequentially selected in all cycles, and the corresponding attitude angle progressive change curves are time-normalized to generate standard periodic deviation waveform templates. Then, the dynamic time warping (DTW) method is used to measure the degree of time offset matching between all cycles, and the waveform point with the largest deviation is taken as the progressive deviation point. The energy value of the integral of the overall time axis of the deviation waveform and the degree of periodic deviation are used as two target indicators, marked as the behavior deviation energy intensity sequence. Finally, the trajectory behavior progressive deviation learning data is output. The main fields include feature vectors such as trajectory segment identification code, aggravation frequency, amplitude accumulation energy, periodic progressive delay rate and waveform structure stability, which provide change law classification for the subsequent optimization model update of the trajectory tracking control system. The processing process is based on the open-loop trajectory command and feedback error statistical method. 88 significant deviation segments are formed in 1600 cycle training data, with an average of 7.2 deviation points detected per cycle. The total processing time is 14 seconds, and the error fit goodness R² reaches 0.94, which has a high trajectory recognition accuracy and stable monitoring foundation.

[0038] Step S233 includes the following steps: Perform jitter spatial orientation disorder intensity analysis on the trajectory jitter amplitude skew intensity data to obtain jitter spatial orientation disorder intensity data; According to the jitter spatial azimuth disorder intensity data, the spatial azimuth vector angular rate mutation entropy value is analyzed on the progressive mean difference of the attitude angle offset to obtain the azimuth vector angular rate mutation entropy value; The joint angle overshoot probability is calculated based on the jitter spatial orientation disorder intensity data for the joint angle deviation time series increment ratio to obtain the joint angle overshoot probability calculation data; Based on the azimuth vector angular rate mutation entropy value and the joint angle overshoot probability calculation data, the robot trajectory jamming swing amplitude regression analysis is performed to obtain the robot trajectory jamming swing amplitude regression data.

[0039] In the embodiments of the present invention, jitter spatial orientation disorder intensity analysis is performed on the trajectory jitter amplitude skewness intensity data. When performing this analysis, first, the jitter amplitude skewness intensity data is mapped to the displacement deviation direction of the end effector of the robotic arm in three-dimensional space, and the deviation signals and their corresponding amplitude state intensity values in the three axes of X, Y, and Z are extracted, which are regarded as the sampling points of the three-dimensional vector field. For each sampling point in the continuous period, the spatial direction where it is located is divided into unit spaces, with a cube side length of 10 millimeters as the unit grid, and the number of deviation vectors and intensity distribution within each grid are counted. For each grid, the disorder intensity index is calculated based on the vector direction distribution within it, that is, by traversing all the deviation vectors inside the grid, calculating the absolute value of the angle between them and the average direction within the grid, and statistically calculating the standard deviation of the angles as the disorder intensity value of the grid. The disorder intensity values of all grids constitute the jitter spatial orientation disorder intensity data structure, with fields including the grid center coordinates, the number of vectors, the average amplitude, and the standard deviation of the angles, accurate to three decimal places, and output as a structured table. Based on the jitter spatial orientation disorder intensity data, spatial orientation vector angular rate mutation entropy value analysis is performed on the attitude angle offset progressive mean difference. First, the attitude angle offset progressive mean difference data is temporally and spatially corresponding and matched with the spatial disorder intensity data, and the disorder intensity value of the corresponding sampling grid is extracted as the weight for each sampling period. Subsequently, for the average attitude angle offset vector in this period, the mutation rates of its angular rate in the three axes of space are calculated, that is, by calculating the direction change rate between adjacent timestamps and weighting the disorder intensity value of this period to obtain the weighted angular rate mutation sequence. Shannon entropy calculation is performed on this sequence, that is, by statistically calculating the probability of the angular rate mutation value interval distribution and then calculating the entropy value, which is used to reflect the complexity of the change in the attitude angle offset direction. The output fields include the period number, the average disorder intensity, the angular rate mutation entropy value, the maximum direction change interval, etc. Based on the jitter spatial orientation disorder intensity data, the joint angle overshoot probability calculation (that is, the angle change exceeds the target or expectation) is performed on the joint angle deviation time series increment ratio. This step first segments the joint angle deviation time series increment ratio data according to each joint number, and each segment contains 20 consecutive period data. For each joint data segment, the event frequency that its increment ratio exceeds the critical value (set to 0.8, that is, the increment reaches 80% of the maximum ratio) is statistically calculated, and the conditional probability is calculated in combination with the spatial disorder intensity value of the corresponding period, that is, calculating the probability of angle increment overshoot occurring in the high spatial disorder grid. By smoothing the probability in the time dimension with a sliding window (window length 5 periods), a joint angle overshoot probability curve is formed, and the output fields are the joint number, the period number, and the overshoot probability value. Based on the azimuth vector angular rate mutation entropy value and the joint angle overshoot probability calculation data, a robotic arm trajectory jitter amplitude regression analysis is performed. This analysis uses the multiple generalized linear regression method, taking the angular rate mutation entropy value and the overshoot probability of each joint as independent variables and the actual jitter amplitude of the end effector as the dependent variable to construct a regression model.The regression process uses a 30-trajectory execution cycle as the sample window to iteratively estimate the regression coefficients. The ridge regression method is used to handle the collinearity problem, and the regularization parameter is set to 0.01 to ensure stability. The regression results record the coefficients of each independent variable, the standard error, the p-value, and the residual variance. The regression model is evaluated by ten-fold cross-validation, and the average prediction error is controlled within 0.1 mm. The finally output regression data of the robotic arm trajectory jitter swing includes fields such as cycle number, independent variable value, independent variable coefficient, predicted value, residual value, R-squared value, etc.

[0040] In another embodiment, the trajectory jitter amplitude skew intensity data is decomposed into X, Y, and Z direction components according to the three-dimensional space coordinates, and the sampling frequency is 200Hz. The amplitude data in each direction is Hilbert transformed to obtain the instantaneous phase, and the phase resolution is 0.01 radians. The phase difference between adjacent sampling points is calculated to construct a phase difference sequence. The phase difference sequence is autocorrelated, and the delay time is from 1 to 100 sampling points. The first zero crossing point position of the autocorrelation function is extracted as the correlation length. The mutual correlation coefficient matrix of the amplitudes in the three directions is calculated. The mutual correlation matrix is singular value decomposition is performed, and the maximum three singular values are taken. The correlation length, singular value, and phase difference statistics are combined to form the jitter space orientation disorder intensity data. The attitude angle offset progressive mean difference data is resampled to 200Hz and aligned with the jitter space orientation disorder intensity data. The angular velocity sequence of the six joints is calculated, and the five-point central difference method is used. The angular velocity sequence is segmented, and each segment is 1000 sampling points long. Detect angular velocity mutation points in each segment, and set the mutation threshold to 2.5 times the local standard deviation. Count the number of mutation points and their time distribution in each segment. Calculate the angular velocity change at the mutation point, and calculate the weighted entropy value in combination with the jitter space orientation disorder intensity data. The weight coefficient is determined by the singular value of the disorder intensity. Integrate the entropy value sequence, mutation statistics, and angular velocity change characteristics of each segment to obtain the azimuth vector angular velocity mutation entropy value. Segment the joint angle deviation time series increment ratio data, and the segment length is 2 seconds. Set the angle threshold boundary in each segment, with the upper limit being the mean plus 2 times the standard deviation and the lower limit being the mean minus 2 times the standard deviation. Count the data points that exceed the threshold boundary and calculate the overshoot time ratio. Use the jitter space orientation disorder intensity data to weight the overshoot event, and the weight is proportional to the disorder intensity. Perform probability density estimation on the weighted overshoot sequence, using the kernel density estimation method with a bandwidth of 0.1. Calculate the overshoot probability distribution parameters of each joint, including mean, variance, skewness, and kurtosis. Combine the probability distribution parameters with the overshoot statistics to form the joint angle overshoot probability calculation data. Use the azimuth vector angular rate mutation entropy value and the joint angle overshoot probability calculation data as input features to construct a support vector regression model. Use the radial basis kernel function, set the kernel parameter gamma to 0.01, and set the penalty factor C to 10. Divide the data set in chronological order, with the first 80% used for training and the last 20% for testing. Normalize the features to a normalization interval of [-1,1]. Train the support vector regression model and use grid search to determine the optimal hyperparameters. Make predictions on the test set and calculate the prediction error statistics. Integrate the prediction results, model parameters, and error statistics into the robot trajectory jamming amplitude regression data.

[0041] Step S24 includes the following steps: Step S241: performing robot trajectory error clustering processing on the trajectory behavior progressive deviation learning data to obtain robot trajectory error clustering data; Step S242: Based on the robotic arm trajectory error clustering data, perform a control logic structure reconstruction process on the robotic arm trajectory tracking execution program to generate control program logic structure reconstruction data; Step S243: Impose a control logic nesting depth constraint on the control program logic structure reconstruction data to obtain program control logic depth constraint reconstruction data; Step S244: According to the program control logic depth constraint reconstruction data, perform trajectory tracking control iteration optimization on the robotic arm trajectory tracking execution program to obtain a trajectory tracking execution optimization program.

[0042] In the embodiments of the present invention, when performing robotic arm trajectory error clustering processing on the trajectory behavior progressive deviation learning data, first, the trajectory behavior progressive deviation learning data is segmented according to time windows, and each segment contains trajectory data of 20 control cycles. Three types of error metrics, namely spatial displacement error, attitude angle offset, and end effector offset, are extracted for each cycle. After normalizing the error amounts, a sample set is constructed in the form of a three-dimensional error vector, and the Density Peak Clustering algorithm is used for unsupervised classification. The error trends are divided into three categories: stable deviation type, periodic deviation type, and mutation deviation type. When clustering, the local density threshold is set to 0.4, and the distance threshold is set to 0.6. Finally, a set of labeled error clustering data is obtained, and each category contains its clustering center coordinates, the data indices it belongs to, and the average offset vector, which is used to characterize the deviation distribution pattern of the trajectory behavior. When performing control logic structure reconstruction processing on the robotic arm trajectory tracking execution program based on the robotic arm trajectory error clustering data, the trajectory execution code segments corresponding to each type of error are extracted, and the control logic paths with a frequency higher than 0.8 in the stable deviation type are identified. The conditional structures with frequent loops are expanded and replaced with a branch structure driven by a state transition matrix. The control logic corresponding to the periodic deviation type clustering uses the periodic conditional jump optimization technique to optimize the original time-based decision formula into a trajectory rhythm conditional control structure based on a phase-locker. The code segments corresponding to the mutation deviation type clustering are subjected to enhanced abnormal jump capture processing, and high-priority error event interruption response logic is inserted to generate structured control program logic structure reconstruction data, which includes a logic tree structure, a conditional path table, and a trigger abnormal processing logic mapping table. When performing control logic nested depth constraint on the control program logic structure reconstruction data, all multi-level nested conditional structures in all control logic trees are analyzed, and three-dimensional parameters, namely the number of logic levels, the conditional complexity of each level, and the number of nested paths, are extracted. The maximum number of nested levels is set to 5. When it is detected that there is a situation where the number of nested paths in the control structure exceeds 5 levels, the control segment is automatically split into two sub-process modules, and the original deep nested logic is reconstructed into a linear structure in the form of state jumps using the logic callback mechanism, and the number of nested levels is reduced to within 3. At the same time, a Boolean expression simplification strategy is introduced, and the Karnaugh map method is used to simplify the complex conditional judgment logic to ensure that the total number of control logic paths does not exceed 128. After processing, program control logic depth constraint reconstruction data is formed, and the data format includes control block identification, the reconstructed path depth value, and the corresponding trigger response structure description. When performing trajectory tracking control iterative optimization on the robotic arm trajectory tracking execution program according to the program control logic depth constraint reconstruction data, a control logic version management chain is established, and an independent version identifier is generated for the control program after each logic structure adjustment.For each version, 100 typical trajectories (such as circular trajectories, figure-eight trajectories, and right-angle paths) are executed using an actual robotic arm to obtain real-time error data, and a trajectory error feedback matrix is constructed. By comparing the difference between the current version's trajectory error feedback mean and that of the previous version, the error convergence trend is extracted. If the error reduction amplitude of the current version is less than 5%, the control weight adjustment module is activated to fine-tune the proportional gain term and integral feedback term in the current control program, with the step size set to 0.01, ensuring that the control program achieves the minimum step size optimization in the convergence direction. The finally output trajectory tracking execution optimization program includes the optimized version number, the error convergence rate index, and the updated control logic structure mapping table.

[0043] Step S3 includes the following steps: Step S31: Perform static analysis on the program code of the trajectory tracking execution optimization program to obtain static analysis data of the program code; Step S32: Perform functional unit integration testing on the trajectory tracking execution optimization program based on the static analysis data of the program code to obtain functional unit integration testing data; Step S33: Design the robotic arm trajectory tracking control firmware based on the functional unit integration testing data to obtain the robotic arm trajectory tracking control firmware; Step S34: Embed the robotic arm trajectory tracking control firmware into the robotic arm control center to execute the robotic arm trajectory tracking control.

[0044] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: Perform static analysis on the program code of the trajectory tracking execution optimization program to obtain static analysis data of the program code; In the embodiment of the present invention, when performing static analysis on the program code of the trajectory tracking execution optimization program, a method combining syntax tree parsing and data flow analysis is adopted to analyze the source code of the optimization control program. First, a syntax parser (such as Clang AST) is used to construct an abstract syntax tree for the control program, and the function call graph, variable definition-use chain, and semantic context relationship of each control module are extracted. Then, a program control flow graph (CFG) and a data flow graph (DFG) are constructed to analyze potential problems in the program, such as variable escape paths, uninitialized variables, and insufficient branch jump coverage. A static analysis tool is used for complexity measurement, and the cyclomatic complexity is adopted to evaluate the number of control paths of each function. If it exceeds 20, the function is marked as a high-complexity function. In the example, when analyzing the program module containing the main loop of trajectory control, it is found that there are 3 conditional jump statements with incomplete path coverage, and there is a risk of uninitialized use of the integral error variable in a certain joint PID controller. The finally generated static analysis data of the program code includes a control path graph, a variable usage table, a code complexity report, and a list of potential error warnings.

[0045] Step S32: Perform functional unit integration testing on the trajectory tracking execution optimization program according to the static analysis data of the program code to obtain functional unit integration test data; In the embodiment of the present invention, when performing functional unit integration testing on the trajectory tracking execution optimization program according to the static analysis data of the program code, a functional unit-level verification mechanism based on the test-driven development framework is adopted. First, according to the function call dependency graph generated by static analysis, a functional unit test list is constructed, and independent test modules are constructed for each main control function (such as trajectory generation function, error feedback processing function, position-velocity-acceleration command fusion function). An industrial control-level simulation platform (such as Matlab Simulink with TargetLink) is used to build a simulation environment, and standard trajectory data is injected for closed-loop execution testing. The test parameters include that the trajectory sampling period is set to 5 ms, the maximum command position step is 0.02 rad, and the error injection range is controlled within ±1 mm. Boundary value testing, abnormal input testing, and instruction jump testing are set for each functional unit to verify its robustness. The finally generated functional unit integration test data includes the input-output consistency verification results of each function, the logical path coverage rate (required to reach more than 95%), the execution delay statistics (the maximum shall not exceed 1 ms), and the output response error statistics table.

[0046] Step S33: Design the manipulator trajectory tracking control firmware for the trajectory tracking execution optimization program based on the functional unit integration test data to obtain the manipulator trajectory tracking control firmware; In the embodiment of the present invention, when designing the firmware for robotic arm trajectory tracking control of the trajectory tracking execution optimization program based on the functional unit integration test data, a real-time embedded system development platform (such as TI C2000 series or STM32H7) is used to complete the firmware design process. The tested trajectory control program is compiled into target instruction codes (.hex format or.bin format) using a cross-compiler. During the compilation process, the interrupt priorities are configured, the trajectory control main loop is allocated to the periodic interrupt service function, and the high-frequency data acquisition function is set to the DMA drive mode to reduce the CPU load. For different joint channels, independent timer resources and PWM output modules are allocated, and the system clock is configured in the startup code, with the main frequency set to 400 MHz to meet the 5 ms control cycle requirement. A firmware interface protocol is designed to communicate with the host computer via RS485 or CAN bus, and the communication frame structures for trajectory update instructions, joint status query, and emergency stop control are defined. The finally output firmware for robotic arm trajectory tracking control is a binary file that can be burned into the target controller, accompanied by a hardware resource mapping description table and a register initialization table.

[0047] Step S34: Embed the firmware for robotic arm trajectory tracking control into the robotic arm control center to perform robotic arm trajectory tracking control.

[0048] In the embodiment of the present invention, when embedding the firmware for robotic arm trajectory tracking control into the robotic arm control center to perform robotic arm trajectory tracking control, an industrial programming device is used to write the generated firmware into the FLASH storage area of the control main board. After programming is completed, the control system is powered on and started, and software and hardware joint debugging is carried out. The controller status is monitored online through the JTAG interface, and the joint drive output signals are captured using an oscilloscope to verify whether the trajectory control cycle is strictly 5 ms, whether the PWM output waveform is continuous, and whether the current command change is smooth without jitter. A three-dimensional position tracker (such as the OptiTrack system) is used to monitor the trajectory execution of the robotic arm end in real time, and the error is compared and analyzed with the theoretical trajectory, and the trajectory execution error of each joint is recorded, requiring that the average deviation is not greater than 1.5 mm and the maximum deviation does not exceed 2.5 mm. After debugging is completed, the firmware configuration data is written into the EEPROM as the startup configuration file, and the robotic arm control center enters the working state and starts to perform trajectory tracking control.

[0049] The present invention also provides a robotic arm trajectory tracking control system for performing the robotic arm trajectory tracking control method as described above. The robotic arm trajectory tracking control system includes: A trajectory jitter perception analysis module, which is used to obtain the operation log of the robotic arm and the robotic arm trajectory tracking execution program; extract the trajectory motion deviation associated with the usage period from the operation log of the robotic arm to obtain the periodic trajectory motion deviation data; perform robotic arm motion trajectory jitter perception record analysis based on the periodic trajectory motion deviation data to obtain the trajectory deviation jitter perception record data; A behavior deviation feature learning module, which is used to perform robotic arm trajectory behavior deviation feature learning according to the trajectory deviation jitter perception record data to obtain the trajectory behavior progressive deviation learning data; perform iterative optimization of the trajectory tracking control on the robotic arm trajectory tracking execution program according to the trajectory behavior progressive deviation learning data to obtain the trajectory tracking execution optimization program; A tracking control firmware design module, which is used to design the robotic arm trajectory tracking control firmware for the trajectory tracking execution optimization program to obtain the robotic arm trajectory tracking control firmware; embed the robotic arm trajectory tracking control firmware into the robotic arm control center to execute the robotic arm trajectory tracking control.

[0050] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. A manipulator trajectory tracking control method, characterized in that It includes the following steps: Step S1: Obtain the robotic arm operation log and the robotic arm trajectory tracking execution program; Extract the trajectory motion deviation associated with the usage period from the robotic arm operation log to obtain the periodic trajectory motion deviation data; perform analysis on the perception record of the robotic arm motion trajectory jitter based on the periodic trajectory motion deviation data to obtain the trajectory deviation jitter perception record data; Step S2: Perform learning on the robotic arm trajectory behavior deviation characteristics according to the trajectory deviation jitter perception record data to obtain the trajectory behavior progressive deviation learning data; perform iterative optimization of the trajectory tracking control on the robotic arm trajectory tracking execution program according to the trajectory behavior progressive deviation learning data to obtain the trajectory tracking execution optimization program; Step S3: Design the robotic arm trajectory tracking control firmware for the trajectory tracking execution optimization program to obtain the robotic arm trajectory tracking control firmware; embed the robotic arm trajectory tracking control firmware into the robotic arm control center to execute the robotic arm trajectory tracking control.

2. The robotic arm trajectory tracking control method according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Obtain the robotic arm operation log and the robotic arm trajectory tracking execution code; Step S12: Clean the data of the robotic arm operation log to obtain the cleaned robotic arm operation log; Step S13: Extract the trajectory motion deviation associated with the usage period from the cleaned robotic arm operation log to obtain the periodic trajectory motion deviation data; Step S14: Perform analysis on the perception record of the robotic arm motion trajectory jitter on the cleaned robotic arm operation log based on the periodic trajectory motion deviation data to obtain the trajectory deviation jitter perception record data.

3. The robotic arm trajectory tracking control method according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Analyze the skewness intensity of the trajectory jitter amplitude change for the trajectory deviation jitter perception record data to obtain the trajectory jitter amplitude skewness intensity data; Step S22: Perform progressive analysis of the spatial deviation amount on the periodic trajectory motion deviation data to obtain the progressive data of the deviation amount between trajectory motions; Step S23: Perform learning on the robotic arm trajectory behavior progressive deviation characteristics according to the trajectory jitter amplitude skewness intensity data and the progressive data of the deviation amount between trajectory motions to obtain the trajectory behavior progressive deviation learning data; Step S24: Perform iterative optimization of the trajectory tracking control on the robotic arm trajectory tracking execution program according to the trajectory behavior progressive deviation learning data to obtain the trajectory tracking execution optimization program.

4. The robotic arm trajectory tracking control method according to claim 3, characterized in that Step S21 includes the following steps: Step S211: Perform time-frequency domain conversion on the trajectory deviation jitter perception record data to generate a deviation jitter time-frequency domain diagram; Step S212: Identify the skewness of the distribution of jitter frequency mutation points for the deviation jitter time-frequency domain diagram to obtain the skewness data of the distribution of jitter frequency mutation points; Step S213: Calculate the variance of the increasing slope of the mutation points according to the skewness data of the distribution of jitter frequency mutation points to obtain the variance of the increasing slope of the frequency mutation points; Step S214: Perform convergence constraint logarithmic fitting processing on the deviation jitter time-frequency domain diagram based on the variance of the increasing slope of the frequency mutation points to obtain the convergence logarithmic fitting data of the frequency mutation points; Step S215: Analyze the skewness intensity of the trajectory jitter amplitude change according to the convergence logarithmic fitting data of the frequency mutation points to obtain the trajectory jitter amplitude skewness intensity data.

5. The robotic arm trajectory tracking control method according to claim 4, wherein Step S214 includes the following steps: Perform an increasing segmented exponential transformation on the frequency domain curve of the mutation points of the deviation jitter time series frequency domain diagram based on the variance of the increasing slope of the frequency mutation points to obtain the increasing segmented exponential of the mutation point curve; Perform an increasing segmented exponential ratio convergence process on the increasing segmented exponential of the mutation point curve to obtain the increasing segmented exponential convergence ratio; Perform a convergence constraint logarithmic fitting process on the deviation jitter time series frequency domain diagram according to the increasing segmented exponential convergence ratio to obtain the frequency mutation point convergence logarithmic fitting data.

6. The robotic arm trajectory tracking control method according to claim 3, wherein Step S23 includes the following steps: Step S231: Calculate the progressive mean difference of the attitude angle offset for the progressive data of the deviation amount between trajectory movements to obtain the progressive mean difference of the attitude angle offset; Step S232: Calculate the ratio of the time series increment of the joint angle deviation for the progressive data of the deviation amount between trajectory movements to obtain the ratio of the time series increment of the joint angle deviation; Step S233: Perform a robotic arm trajectory jitter amplitude regression analysis on the progressive mean difference of the attitude angle offset and the ratio of the time series increment of the joint angle deviation based on the trajectory jitter amplitude skewness intensity data to obtain the robotic arm trajectory jitter amplitude regression data; Step S234: Perform a transfer learning of the repeated trajectory jitter amplitude on the robotic arm trajectory jitter amplitude regression data to obtain the repeated trajectory jitter amplitude learning data; Step S235: Perform a progressive deviation feature learning of the robotic arm trajectory behavior based on the repeated trajectory jitter amplitude learning data to obtain the progressive deviation learning data of the trajectory behavior.

7. The robotic arm trajectory tracking control method according to claim 6, characterized in that Step S233 includes the following steps: Analyze the disorder intensity of the jitter space orientation of the trajectory jitter amplitude skewness intensity data to obtain the disorder intensity data of the jitter space orientation; Perform a spatial orientation vector angular rate mutation entropy value analysis on the progressive mean difference of the attitude angle offset based on the disorder intensity data of the jitter space orientation to obtain the azimuth vector angular rate mutation entropy value; Perform a joint angle overshoot probability calculation on the ratio of the time series increment of the joint angle deviation based on the disorder intensity data of the jitter space orientation to obtain the joint angle overshoot probability calculation data; Perform a robotic arm trajectory jitter amplitude regression analysis based on the azimuth vector angular rate mutation entropy value and the joint angle overshoot probability calculation data to obtain the robotic arm trajectory jitter amplitude regression data.

8. The robotic arm trajectory tracking control method according to claim 3, wherein, Step S24 includes the following steps: Step S241: Cluster the robotic arm trajectory error for the progressive deviation learning data of the trajectory behavior to obtain the robotic arm trajectory error clustering data; Step S242: Reconstruct the control logic structure of the robotic arm trajectory tracking execution program based on the robotic arm trajectory error clustering data to generate the reconstructed data of the control program logic structure; Step S243: Constrain the nested depth of the control logic for the reconstructed data of the control program logic structure to obtain the reconstructed data of the program control logic depth constraint; Step S244: Iteratively optimize the trajectory tracking control of the robotic arm trajectory tracking execution program according to the reconstructed data of the program control logic depth constraint to obtain the optimized program for the trajectory tracking execution.

9. The robotic arm trajectory tracking control method according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Perform a static analysis of the program code for the optimized program of the trajectory tracking execution to obtain the static analysis data of the program code; Step S32: Perform a functional unit integration test on the optimized program of the trajectory tracking execution according to the static analysis data of the program code to obtain the functional unit integration test data; Step S33: Based on the functional unit integration test data, perform the firmware design of the robotic arm trajectory tracking control for the trajectory tracking execution optimization program to obtain the robotic arm trajectory tracking control firmware; Step S34: Embed the robotic arm trajectory tracking control firmware into the robotic arm control center to perform the robotic arm trajectory tracking control.

10. A robotic arm trajectory tracking control system, characterized in that, For executing the robotic arm trajectory tracking control method as claimed in claim 1, the robotic arm trajectory tracking control system comprises: A trajectory jitter perception analysis module, configured to obtain the robotic arm operation log and the robotic arm trajectory tracking execution program; extract the trajectory motion deviation associated with the usage cycle from the robotic arm operation log to obtain the periodic trajectory motion deviation data; perform the perception record analysis of the robotic arm motion trajectory jitter based on the periodic trajectory motion deviation data to obtain the trajectory deviation jitter perception record data; A behavior deviation feature learning module, configured to perform the trajectory behavior deviation feature learning of the robotic arm according to the trajectory deviation jitter perception record data to obtain the trajectory behavior progressive deviation learning data; perform the iterative optimization of the trajectory tracking control on the robotic arm trajectory tracking execution program according to the trajectory behavior progressive deviation learning data to obtain the trajectory tracking execution optimization program; A tracking control firmware design module, configured to perform the firmware design of the robotic arm trajectory tracking control for the trajectory tracking execution optimization program to obtain the robotic arm trajectory tracking control firmware; embed the robotic arm trajectory tracking control firmware into the robotic arm control center to perform the robotic arm trajectory tracking control.

Images