Six-degree-of-freedom mechanical arm control method and system

Abstract

The invention relates to the technical field of six-degree-of-freedom mechanical arm control, in particular to a six-degree-of-freedom mechanical arm control method and system. The method comprises the following steps: obtaining an assembly schematic diagram, carrying out multi-machine cooperation process analysis, and collecting the vibration frequency of a vehicle body structure by using an acoustic emission sensor; thirdly, vibration impact force disorder of the connection interface is simulated through the vibration frequency, pre-tightening force misalignment regression analysis is carried out, and therefore fatigue fracture failure of the connection structure is predicted; then, based on regression data and a prediction result, torque-angle compound control of the six-degree-of-freedom mechanical arm is implemented, and iteration is carried out; and finally, control firmware is designed and embedded into the mechanical arm, and precise control is executed. Through optimization processing of the six-degree-of-freedom mechanical arm control technology, the six-degree-of-freedom mechanical arm control technology is more perfect.

Description

Technical Field

[0001] The present invention relates to the technical field of six-degree-of-freedom robotic arm control, and particularly to a six-degree-of-freedom robotic arm control method and system. Background Art

[0002] As an advanced robot technology, the six-degree-of-freedom robotic arm has become one of the key devices on the automated production line due to its high degrees of freedom, high flexibility, and programmability. The six-degree-of-freedom robotic arm can achieve precise multi-directional operations in space, simulate the movements of a human arm, and can perform various tasks such as precise positioning, handling, welding, and assembly. In practical applications, the control of the six-degree-of-freedom robotic arm faces many technical challenges. First of all, the operation of the robotic arm often involves complex mechanical calculations. Especially in the process of precision assembly, how to achieve precise cooperation and force control between the robotic arm and the workpiece has become a key issue. Secondly, factors such as vibration, impact, and pre-tightening force that need to be dealt with during the assembly process make the accuracy and stability of the robotic arm control system the core of improving production efficiency. However, there is a problem in a traditional six-degree-of-freedom robotic arm control method that the analysis of the pre-tightening force misalignment of the assembly connection structure of automotive parts is inaccurate, resulting in large control errors of the six-degree-of-freedom robotic arm. Summary of the Invention

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

[0004] To achieve the above object, a six-degree-of-freedom robotic arm control method, the method includes the following steps: Step S1: Obtain the assembly schematic diagram of automotive parts; perform a multi-robot cooperation process analysis on the assembly schematic diagram of automotive parts to obtain the multi-robot cooperation process for parts assembly; based on the multi-robot cooperation process for parts assembly, and through the acoustic emission sensor installed on the six-degree-of-freedom robotic arm, collect the vibration frequencies of the multi-robot assembly structure to obtain the vibration frequencies of the assembled body structure filled in time series; Step S2: Simulate the disorder of the vibration direction and impact of the assembly connection structure interface based on the vibration frequencies of the assembled body structure filled in time series to obtain the disordered data of the vibration direction impact force of the connection interface; perform a regression analysis of the pre-tightening force misalignment based on the disordered data of the vibration direction impact force of the connection interface to obtain the regression data of the pre-tightening force misalignment; predict the fatigue fracture failure of the connection structure according to the regression data of the pre-tightening force misalignment to obtain the prediction data of the fatigue fracture of the connection structure; Step S3: Perform torque-angle composite control of the six-degree-of-freedom robotic arm according to the regression data of the pre-tightening force misalignment and the prediction data of the fatigue fracture of the connection structure to obtain the iterative data of torque-angle control; Step S4: Design the six-degree-of-freedom robotic arm control firmware based on the torque-angle control iterative data to obtain the torque-angle control firmware; embed the torque-angle control firmware into the six-degree-of-freedom robotic arm to perform the six-degree-of-freedom robotic arm control.

[0005] Preferably, step S1 includes the following steps: Step S11: Obtain the assembly schematic diagram of automotive parts. Step S12: Analyze the multi-robot cooperation process of the assembly schematic diagram of automotive parts to obtain the multi-robot cooperation process for parts assembly. Step S13: Based on the multi-robot cooperation process for parts assembly, collect the vibration frequencies of the multi-robot assembled body structure through the acoustic emission sensors installed on the six-degree-of-freedom robotic arm to obtain the vibration frequencies of the multi-robot assembled body structure. Step S14: Insert time series tags into the vibration frequencies of the multi-robot assembled body structure to obtain the time series vibration frequencies of the assembled structure. Step S15: Fill in the missing values of the time series vibration frequencies of the assembled structure to obtain the time series filled vibration frequencies of the assembled body structure.

[0006] Preferably, step S2 includes the following steps: Step S21: Analyze the asymmetry of the time series filled vibration frequencies of the assembled body structure to obtain the asymmetric dynamic frequencies of the assembled body structure. Step S22: Simulate the disorder of the vibration azimuth impact of the assembly connection structure interface based on the assembly schematic diagram of automotive parts and the asymmetric dynamic frequencies of the assembled body structure to obtain the disordered data of the vibration azimuth impact force of the connection interface. Step S23: Conduct a regression analysis of the misalignment of the connection pre-tightening force based on the assembly schematic diagram of automotive parts and the disordered data of the vibration azimuth impact force of the connection interface to obtain the regression data of the misalignment of the connection pre-tightening force. Step S24: Predict the fatigue fracture failure of the connection structure based on the disordered data of the vibration azimuth impact force of the connection interface and the regression data of the misalignment of the connection pre-tightening force to obtain the prediction data of the fatigue fracture of the connection structure.

[0007] Preferably, step S22 includes the following steps: Step S221: Extract the morphological structure of the connecting parts from the assembly schematic diagram of automotive parts to obtain the morphological structure data of the connecting parts. Step S222: Conduct a coupling analysis of the vibration transfer of the connecting parts on the morphological structure data of the connecting parts according to the asymmetric dynamic frequencies of the assembled body structure to obtain the coupling data of the vibration transfer of the connecting parts. Step S223: Fit the random process of the vibration azimuth of the coupling data of the vibration transfer of the connecting parts to obtain the random process data of the vibration azimuth of the connecting parts. Step S224: Based on the random process data of the vibration orientation of the connecting piece, the vibration transfer coupling data of the connecting piece, and the asymmetric dynamic frequency of the assembled body structure, perform superposition fitting of the impact force in the connection interface orientation to obtain the superposition data of the impact force in the connection interface orientation; Step S225: According to the superposition data of the impact force in the connection interface orientation, perform simulation of the disorder of the impact in the vibration orientation of the assembled connection structure interface to obtain the disorder data of the impact in the vibration orientation of the connection interface.

[0008] Preferably, step S224 includes the following steps: Calculate the ratio of the acceleration of the frequency response of the vibration transfer coupling data of the connecting piece to obtain the ratio of the acceleration of the vibration frequency of the connecting piece; Based on the random process data of the vibration orientation of the connecting piece, perform coupling processing of the multi-orientation frequency diffusion effect on the ratio of the acceleration of the vibration frequency of the connecting piece to obtain the coupling data of the orientation frequency diffusion effect; Perform interpolation processing of the diffusion radius of the frequency vibration energy intensity on the coupling data of the orientation frequency diffusion effect to obtain the interpolation data of the diffusion radius of the frequency intensity; According to the interpolation data of the diffusion radius of the frequency intensity and the asymmetric dynamic frequency of the assembled body structure, identify the equal gradient of the spatial energy density vector between the connection interfaces to obtain the equal gradient of the spatial energy density vector; Based on the equal gradient of the spatial energy density vector, perform superposition fitting of the impact force in the connection interface orientation to obtain the superposition data of the impact force in the connection interface orientation.

[0009] Preferably, step S23 includes the following steps: Step S231: Extract the theoretical pre-tightening force of the assembly connection torque based on the assembly schematic diagram of the automotive parts to obtain the theoretical pre-tightening force of the assembly connection torque; Step S232: Perform multi-scale analysis of the impact force in the vibration orientation of the disorder data of the impact force in the connection interface vibration orientation to obtain the multi-scale data of the disorder impact force in the orientation vibration; Step S233: Perform analysis of the difference in the skewness of the impact force fluctuation in the spatial dimension on the multi-scale data of the disorder impact force in the orientation vibration to obtain the difference data of the skewness of the spatial impact force fluctuation; Step S234: According to the difference data of the skewness of the spatial impact force fluctuation, perform simulation identification of the pre-tightening force interference misalignment gradient on the theoretical pre-tightening force of the assembly connection torque to obtain the pre-tightening force interference misalignment gradient data; Step S235: Perform regression analysis of the misalignment of the connection pre-tightening force on the pre-tightening force interference misalignment gradient data to obtain the regression data of the misalignment of the connection pre-tightening force.

[0010] Preferably, step S234 includes the following steps: Perform equal ratio calculation of the difference in the skewness and kurtosis of the spatial orientation on the difference data of the skewness of the spatial impact force fluctuation to obtain the equal ratio data of the difference in the skewness and kurtosis of the orientation; Based on the azimuth skewness kurtosis difference ratio data, a simulation evaluation of the pre-tightening friction slip amount of the theoretical assembly connection torque pre-tightening force is carried out to obtain the pre-tightening friction slip amount data; According to the pre-tightening friction slip amount data and the azimuth skewness kurtosis difference ratio data, an analysis of the local pre-tightening force modal mismatch distribution of the connection surface is carried out to obtain the local pre-tightening force mismatch distribution data; Based on the local pre-tightening force mismatch distribution data, a simulation identification of the pre-tightening force interference misalignment gradient is carried out to obtain the pre-tightening force interference misalignment gradient data.

[0011] Preferably, step S3 includes the following steps: Step S31: Perform convolution processing on the connection pre-tightening force misalignment regression data to obtain the connection pre-tightening force misalignment convolution data; Step S32: Perform feature structure analysis on the connection structure fatigue fracture prediction data to obtain the structure fatigue fracture feature data; Step S33: According to the connection pre-tightening force misalignment convolution data and the structure fatigue fracture feature data, perform torque-angle composite control of the six-degree-of-freedom robotic arm to generate torque-angle composite control data; Step S34: Perform iterative learning on the torque-angle composite control data to obtain the torque-angle control iteration data.

[0012] Preferably, step S33 includes the following steps: Step S331: According to the connection pre-tightening force misalignment convolution data and the structure fatigue fracture feature data, perform timing torque limit matching of the connecting piece to obtain the timing torque limit matching data of the connecting piece; Step S332: Based on the connection pre-tightening force misalignment convolution data, perform perception of the screwing angle correction of the connecting piece to obtain the screwing angle correction perception data; Step S333: According to the timing torque limit matching data of the connecting piece, perform segmented torque application matching on the screwing angle correction perception data to obtain the angle-related torque segmented application data; Step S334: According to the angle-related torque segmented application data, perform torque-angle composite control of the six-degree-of-freedom robotic arm to generate torque-angle composite control data.

[0013] Preferably, the present invention also provides a six-degree-of-freedom robotic arm control system for executing the six-degree-of-freedom robotic arm control method as described above. The six-degree-of-freedom robotic arm control system includes: The structural vibration frequency acquisition module is used to obtain the assembly schematic diagram of automotive parts; perform a multi-machine cooperation process analysis on the assembly schematic diagram of automotive parts to obtain the multi-machine cooperation process for parts assembly; based on the multi-machine cooperation process for parts assembly, and collect the structural vibration frequency of multi-machine assembly through the acoustic emission sensors installed on the six-degree-of-freedom robotic arm to obtain the vibration frequency of the assembly body structure filled in sequence. The connection structure fracture prediction module is used to simulate the disorder of the vibration direction impact at the interface of the assembly connection structure based on the vibration frequency of the assembly body structure filled in sequence to obtain the disorder data of the impact force in the vibration direction at the connection interface; perform a misalignment regression analysis of the connection pre-tightening force based on the disorder data of the impact force in the vibration direction at the connection interface to obtain the misalignment regression data of the connection pre-tightening force; predict the fatigue fracture failure of the connection structure according to the misalignment regression data of the connection pre-tightening force to obtain the fatigue fracture prediction data of the connection structure. The torque-angle composite control module is used to perform torque-angle composite control of the six-degree-of-freedom robotic arm according to the misalignment regression data of the connection pre-tightening force and the fatigue fracture prediction data of the connection structure to obtain the iterative data of torque-angle control. The control firmware design module is used to design the control firmware of the six-degree-of-freedom robotic arm based on the iterative data of torque-angle control to obtain the torque-angle control firmware; embed the torque-angle control firmware into the six-degree-of-freedom robotic arm to execute the control of the six-degree-of-freedom robotic arm.

[0014] The beneficial effects of the present invention are as follows. By obtaining the assembly schematic diagram of automotive parts, it is possible to clearly understand how each part is combined and their mutual relationships. Through the analysis of the multi-robot cooperation process, it can be revealed how different robotic arms divide labor and cooperate during the assembly process to achieve an efficient assembly process. Then, the vibration frequency of the vehicle body structure during the assembly of parts is collected by using the acoustic emission sensor on the six-degree-of-freedom robotic arm, and the vibration frequency of the assembled vehicle body structure is obtained in chronological order. This process not only helps to identify potential assembly problems but also provides vibration data to strongly support subsequent analysis, thereby providing a basis for optimizing the assembly process. Based on the vibration frequency of the assembled vehicle body structure in chronological order, the disorder of the vibration orientation impact of the connection structure interface is simulated. This simulation helps to reveal the disorder of the connection interface under the action of vibration and obtain the disorder data of the impact force. These data provide a basis for further analyzing the performance of the connection structure. Then, by regression analysis of the misalignment of the connection pre-tightening force, the influence of the misalignment of the pre-tightening force on the performance of the connection structure can be further quantified. Combining the regression data of the misalignment of the pre-tightening force, the fatigue fracture failure prediction of the connection structure is carried out, providing a scientific basis for evaluating the long-term reliability and structural safety of the assembly connection, thereby effectively preventing potential fatigue damage and fracture risks. Based on the regression data of the misalignment of the connection pre-tightening force and the fatigue fracture prediction data of the connection structure, the torque-angle composite control of the six-degree-of-freedom robotic arm is carried out. By comprehensively considering the mechanical characteristics and fatigue life of the connection structure, the robotic arm can accurately adjust the torque and angle, thus ensuring that each step in the assembly process meets the predetermined quality standards. By repeatedly iterating and optimizing the control algorithm, not only the assembly accuracy is improved, but also the assembly error caused by improper control of the robotic arm is reduced. The implementation of this control method helps to improve the automation level during the assembly process and provides a more stable control mechanism for the subsequent assembly of products. Based on the torque-angle control iteration data obtained in step S3, a control firmware suitable for the six-degree-of-freedom robotic arm is designed. The design of this firmware should fully consider the dynamic changes and vibration responses during the assembly process to ensure that the robotic arm can perform precise control according to the real-time feedback data. Embedding the designed control firmware into the six-degree-of-freedom robotic arm, the robotic arm can execute the predetermined assembly tasks according to the control instructions of the firmware. By implementing this control scheme, a more efficient and accurate assembly process can be achieved, reducing human operation errors and greatly improving production efficiency, and ultimately providing a more efficient and stable solution for the assembly of automotive parts. Therefore, the present invention makes an optimization treatment for a traditional six-degree-of-freedom robotic arm control method, solves the problem that the traditional six-degree-of-freedom robotic arm control method has inaccurate analysis of the misalignment of the pre-tightening force of the assembly connection structure of automotive parts, resulting in large control errors of the six-degree-of-freedom robotic arm, improves the accuracy of the analysis of the misalignment of the pre-tightening force of the assembly connection structure of automotive parts, and reduces the control error of the six-degree-of-freedom robotic arm. Description of the Drawings

[0015] Figure 1 It is a schematic diagram of the step flow of a six-degree-of-freedom robotic arm control method; Figure 2 It is Figure 1 a detailed implementation step flow schematic diagram of step S2 in Figure 3 It is Figure 1 a detailed implementation step flow schematic diagram of step S3 in Specific implementation manner

[0016] Please refer to Figures 1 to 3 , a six-degree-of-freedom robotic arm control method, the method includes the following steps: Step S1: Obtain the assembly schematic diagram of automotive parts; perform a multi-robot cooperation process analysis on the assembly schematic diagram of automotive parts to obtain the multi-robot cooperation process for parts assembly; based on the multi-robot cooperation process for parts assembly, and collect the vibration frequencies of the multi-robot assembly structure through the acoustic emission sensors installed on the six-degree-of-freedom robotic arm to obtain the vibration frequencies of the assembly body structure filled in sequence; Step S2: Simulate the disorder of the vibration orientation impact of the assembly connection structure interface based on the vibration frequencies of the assembly body structure filled in sequence to obtain the disordered data of the vibration orientation impact force of the connection interface; perform a misalignment regression analysis of the connection pre-tightening force based on the disordered data of the vibration orientation impact force of the connection interface to obtain the misalignment regression data of the connection pre-tightening force; predict the fatigue fracture failure of the connection structure according to the misalignment regression data of the connection pre-tightening force to obtain the fatigue fracture prediction data of the connection structure; Step S3: Perform torque-angle compound control of the six-degree-of-freedom robotic arm according to the misalignment regression data of the connection pre-tightening force and the fatigue fracture prediction data of the connection structure to obtain the torque-angle control iteration data; Step S4: Design the control firmware of the six-degree-of-freedom robotic arm based on the torque-angle control iteration data to obtain the torque-angle control firmware; embed the torque-angle control firmware into the six-degree-of-freedom robotic arm to execute the control of the six-degree-of-freedom robotic arm.

[0017] In the embodiment of the present invention, referring to Figure 1 as described, it is a schematic diagram of the step flow of a six-degree-of-freedom robotic arm control method of the present invention. In this example, the six-degree-of-freedom robotic arm control method includes the following steps: Step S1: Obtain the assembly schematic diagram of automotive parts; perform a multi-robot cooperation process analysis on the assembly schematic diagram of automotive parts to obtain the multi-robot cooperation process for parts assembly; based on the multi-robot cooperation process for parts assembly, and collect the vibration frequencies of the multi-robot assembly structure through the acoustic emission sensors installed on the six-degree-of-freedom robotic arm to obtain the vibration frequencies of the assembly body structure filled in sequence; In the embodiments of the present invention, first, the assembly schematic diagrams of automotive components are extracted from the original design database provided by the vehicle manufacturer. The drawing format is the standard ISO-VDA format, which contains the assembly position information and connection method descriptions of key structural components such as the typical side panels, chassis, longitudinal beams, and doors. After the drawings are imported, an assembly process modeling tool based on the Petri net modeling principle is used for multi-machine cooperation process analysis. This tool realizes the directed graph structure modeling of the multi-station cooperation process by identifying the sequence constraints and resource call paths of multiple six-degree-of-freedom arm operation stations in the assembly task sequence, and outputs the topological structure table of the multi-machine cooperation process for component assembly. In this process table, the assembly part numbers, the station numbers to which they belong, the required assembly actions, the required force control characteristics, and time constraints of each cooperation node are clearly listed, and a task scheduling instruction flow based on the six-degree-of-freedom robotic arm assembly platform is established. At the assembly site, an acoustic emission sensor (AE sensor) is fixedly installed at the end effector part of the six-degree-of-freedom robotic arm, with a frequency response range of 50 kHz to 400 kHz and a sampling frequency set to 1 MHz, and a high-impedance low-noise preamplifier is used to achieve real-time capture of high-sensitivity vibration signals. While the robotic arm performs the assembly action, the acoustic emission signals are synchronously collected, and the duration of each collection is the full cycle length of each assembly cycle, with a time resolution set to 1 millisecond. The collected data is first processed by wavelet packet decomposition, the energy envelope curves of each frequency band are extracted, and unified time sequence tags are inserted with reference to the process flow time line to form the assembly structure time sequence vibration frequency data. The above data is filled in the missing sections using the multi-scale interpolation filling method, and the cubic Hermite spline interpolation is selected as the interpolation method. Finally, the output is an assembly body structure time sequence filled vibration frequency data set containing a complete time series.

[0018] In another embodiment, first, a detailed three-dimensional assembly schematic diagram of automobile parts is obtained, which accurately marks the model, material, geometric dimensions, and assembly connection relationship between each component, including specific connection methods and connection positions such as bolt connection, welding connection, and rivet connection. Then, a multi-machine collaboration process analysis is performed for the three-dimensional assembly schematic diagram. The specific operation is to decompose the entire automobile parts assembly process into several specific assembly task units, such as chassis assembly, body welding, engine installation, interior decoration assembly, etc. For each assembly task unit, the specific operation process of each six-degree-of-freedom robotic arm participating in the collaboration is analyzed in detail, including the motion trajectory of the robotic arm, the type of end effector (for example, welding gun, tightening gun, suction cup), and the sequence and time node of each assembly action. By analyzing the connection relationship of the assembly schematic diagram and the assembly process requirements, the collaborative path of multiple robotic arms in space and the safe area to avoid collision are determined. Then, a highly sensitive acoustic emission sensor is rigidly fixed on the end flange of the selected six-degree-of-freedom robotic arm. The acoustic emission sensor can monitor the weak vibration signal generated by the robotic arm in the process of performing the assembly task in real time. In the process of multi-machine collaborative assembly of the body structure, all the robotic arms participating in the collaboration are started synchronously, and the acoustic emission sensors installed on them are used to collect vibration frequency data. During the collection process, a fixed sampling frequency (for example, 10kHz) and sampling duration are set to ensure that the complete time domain signal of the structural vibration during the assembly process can be captured. For the collected original vibration signal, a time series label insertion operation is performed, specifically, each section of the collected vibration data is accurately timestamped with the specific assembly action and time point being performed at the time, forming vibration frequency data with time series information. For example, the vibration data collected at the 5th second of chassis welding will be labeled "chassis welding-5s". Finally, since some vibration data may be missing due to signal interference or sensor reasons in the actual assembly process, it is necessary to fill the missing values ​​of the vibration frequency data with time series labels. The specific filling method is to use linear interpolation for consecutive missing data points, and perform linear fitting based on the effective vibration frequency values ​​adjacent to the missing point to obtain the estimated value of the missing point. For a small number of discrete missing points, the average value of the adjacent valid data is directly used to fill them, and finally the complete assembled body structure time series filling vibration frequency data with time series information is obtained. This data contains the vibration frequency information of the body structure at different time points and different assembly links in the entire assembly process.

[0019] Step S2: Based on the vibration frequency filled in according to the assembly body structure time sequence, perform a simulation on the disorder of the vibration azimuth impact at the assembly connection structure interface to obtain the disorder data of the vibration azimuth impact force at the connection interface; based on the disorder data of the vibration azimuth impact force at the connection interface, perform a misalignment regression analysis of the connection pre-tightening force to obtain the misalignment regression data of the connection pre-tightening force; according to the misalignment regression data of the connection pre-tightening force, perform a prediction of the fatigue fracture failure of the connection structure to obtain the fatigue fracture prediction data of the connection structure; In the embodiment of the present invention, the vibration frequency data obtained in Step S1 is used to perform the simulation operation of the disorder of the vibration azimuth impact. The non-stationary signal decomposition combined with the fuzzy entropy measure method is adopted to perform the multi-direction impact spectrum expansion on the time-sequence vibration data. First, the assembly structure data is converted into the space coordinate system of each connection interface, and then the local coordinate transformation matrix is established by the direction cosine method to decouple and decompose the direction components of each vector to obtain the impact response vector in each direction; then the fuzzy entropy algorithm is used to evaluate the disorder level of the direction impact intensity, construct the direction-intensity distribution map, and form the disorder data of the vibration azimuth impact force at the connection interface. Next, based on the above impact disorder data, a misalignment regression analysis of the connection pre-tightening force is performed. Based on the bolt connection pre-tightening force value in the theoretical assembly state, combined with the distribution density of the vibration impact intensity in each azimuth in the impact disorder data, a correlation regression process is performed. The partial least squares regression algorithm is used to perform regression modeling on the theoretical pre-tightening force and the torque-stress offset value after being actually affected by the impact, extract the misalignment direction distribution and the influence weight, and output the misalignment regression data table of the connection pre-tightening force, where each connection point corresponds to a set of pre-tightening force offset coefficients and impact response coupling factors on the time axis. Then, the fatigue fracture prediction analysis of the connection structure is carried out by using the above misalignment regression data. The time-domain fatigue cumulative analysis method (Rainflow counting method combined with Palmgren-Miner linear cumulative criterion) is adopted to map the force-vibration history of each connection position into an equivalent stress cycle path, and then map it cycle by cycle with the material fatigue limit S-N curve to predict the lower limit of the fatigue life of this connection point. Finally, the fatigue fracture prediction data of the connection structure is output, and the content includes the remaining fatigue life cycle number, fatigue damage factor, crack initiation position prediction number, etc. of each key connection position.

[0020] Step S3: According to the misalignment regression data of the connection pre-tightening force and the fatigue fracture prediction data of the connection structure, perform the torque-angle composite control of the six-degree-of-freedom robotic arm to obtain the torque-angle control iteration data; In the embodiment of the present invention, first, convolution processing is performed on the connection pre-tightening force misalignment regression data. One-dimensional kernel function sliding convolution is used, and the kernel function is a Gaussian kernel. The kernel width is set to 5 time steps, and the sliding step is 1. The result after convolution is used as the pre-tightening force misalignment convolution data. Subsequently, feature structure extraction is performed on the connection structure fatigue fracture prediction data. The extracted feature dimensions include: the position of the structural fatigue starting point, the crack propagation path length, the crack propagation direction vector, and the fatigue grade factor, forming the structural fatigue fracture feature data. After combining the two types of data, torque-angle composite control generation operations are performed. The cointegration control strategy in the control parameter tuning method is used to construct an angle-torque coupling mapping table. The control reference path is established by the equal-parameter equal-step method to achieve the control of the torque application step size under angle changes. The control parameters are dynamically optimized with the minimum torque difference as the objective function. Finally, a torque-angle composite control data set is formed. Multiple rounds of iterative learning are performed on the above control data set. The algorithm used is an adaptive backpropagation neural network algorithm based on the incremental learning rule. The number of network layers is set to three. The input nodes include the time series pre-tightening force and the fatigue damage grade, and the output is the real-time torque-angle adjustment instruction. During the training process, K-fold cross-validation is used to evaluate the training accuracy. The weight adjustment parameters output in each round of iteration are used to update the control model until the error convergence threshold reaches within 1%. Finally, torque-angle control iterative data is formed for firmware logic control construction.

[0021] In another embodiment, first, convolution processing is performed on the connection pre-tightening force misalignment regression data obtained in step S2. The specific operation is to design a convolution kernel with a specific time window and weight (for example, a Gaussian kernel), and perform a convolution operation on the time series data of the connection pre-tightening force misalignment. The purpose of the convolution operation is to smooth the short-term fluctuations of the pre-tightening force misalignment data and highlight its long-term change trend, obtaining the connection pre-tightening force misalignment convolution data, which reflects the stable trend of the connection pre-tightening force changing over time. Then, feature structure analysis is performed on the connection structure fatigue fracture prediction data obtained in step S2. The specific method is to extract key feature parameters such as the predicted fatigue fracture failure probability or remaining life, and analyze the distribution law and mutual relationship of these parameters among different connection structures, identify the key connection parts that are most likely to occur fatigue fracture, obtaining the structural fatigue fracture feature data, which clearly indicates which connection parts have the highest fatigue risk and failure mode. Next, according to the connection pre-tightening force misalignment convolution data obtained in step S31 and the structural fatigue fracture feature data obtained in step S32, torque-angle composite control of the six-degree-of-freedom robotic arm is performed to generate torque-angle composite control data. The specific control strategy is that for the connection parts with relatively large predicted pre-tightening force misalignment, when the robotic arm performs the tightening operation, adjust the output torque target value. For example, add a compensation amount on the basis of the original target torque, and the size of this compensation amount is proportional to the predicted pre-tightening force loss. At the same time, in order to avoid damage to the connecting parts caused by over-tightening, precise control of the tightening angle is also required. For example, when it is predicted that a certain connecting part is prone to fatigue fracture, its final tightening angle can be appropriately reduced to reduce the stress level inside the connecting part. By comprehensively considering the pre-tightening force misalignment and fatigue fracture risk, a time-varying torque-angle control target sequence, that is, torque-angle composite control data, is generated for each connecting part that needs to be tightened. Finally, iterative learning is performed on the generated torque-angle composite control data. The specific method is that in the actual assembly process, use the torque sensor and angle encoder installed at the end of the robotic arm to collect the actual torque and angle data during the tightening process in real time. Compare these actual data with the torque-angle control target sequence generated in step S33, and calculate the deviation between the two. Then, use the iterative learning control algorithm (for example, model-based iterative learning control, data-based iterative learning control), according to the historical deviation information, correct and optimize the future torque-angle control target sequence to gradually reduce the error between the actual tightening result and the expected target, obtaining the final torque-angle control iterative data, which is a torque-angle control instruction sequence that has been learned and optimized and can better compensate for the pre-tightening force misalignment and reduce the fatigue fracture risk.

[0022] Step S4: Design the six-degree-of-freedom robotic arm control firmware based on the torque-angle control iteration data to obtain the torque-angle control firmware; embed the torque-angle control firmware into the six-degree-of-freedom robotic arm to perform six-degree-of-freedom robotic arm control.

[0023] In the embodiment of the present invention, based on the torque-angle control iteration data obtained in step S3, the design of the six-degree-of-freedom robotic arm control firmware is completed. The control firmware is written based on embedded C language and deployed on the STM32F407VGT6 control chip built into the robotic arm. The firmware program structure includes: an initialization module, a data acquisition module, an attitude control module, a torque adjustment module, and a status feedback module. The modules coordinate and operate with an interrupt trigger mechanism, and the control frequency is 1 kHz. The torque-angle control instruction is transmitted to the servo drive unit in real time through the CAN bus protocol. The driver completes the torque adjustment of each axis motor, and the axis angle change is feedback in real time by an incremental encoder. The closed-loop control strategy is position-force hybrid control. After the control firmware is compiled, it is burned into the internal non-volatile memory of the control chip through the JTAG interface to complete the firmware embedding operation. After the control firmware goes online, it can realize the high-precision assembly task operation of the six-degree-of-freedom robotic arm and maintain high-stability assembly performance in complex force-bearing scenarios of the connection structure.

[0024] Step S1 includes the following steps: Step S11: Obtain the assembly schematic diagram of automotive parts; Step S12: Analyze the multi-robot cooperation process of the assembly schematic diagram of automotive parts to obtain the multi-robot cooperation process of parts assembly; Step S13: Based on the multi-robot cooperation process of parts assembly, and collect the vibration frequency of the multi-robot assembled body structure through the acoustic emission sensor assembled on the six-degree-of-freedom robotic arm to obtain the vibration frequency of the multi-robot assembled body structure; Step S14: Insert time sequence tags into the vibration frequency of the multi-robot assembled body structure to obtain the time sequence vibration frequency of the assembled structure; Step S15: Fill in the missing values of the time sequence vibration frequency of the assembled structure to obtain the time sequence filled vibration frequency of the assembled body structure.

[0025] In the embodiments of the present invention, a detailed three-dimensional assembly schematic diagram of an automotive body-in-white is obtained. This schematic diagram is presented in the form of a CAD model file and precisely includes the geometric shapes, dimensional specifications, and the assembly connection relationships among all the body components (such as the floor panel, side panels, roof panel, doors, front and rear covers, etc.). The connection methods adopted between every two components are clearly marked in the schematic diagram, such as resistance spot welding, laser welding, bolt connection, rivet connection, etc., and the position coordinates of the welding points, the diameters and positions of the bolt holes, as well as the types and quantities of the rivets are detailedly indicated. In addition, this assembly schematic diagram also includes assembly process information, such as the assembly sequence requirements for some key components, as well as specific assembly reference planes and positioning points. For the automotive component assembly schematic diagram obtained in step S11, a multi-robot cooperation process analysis is carried out. The specific operation is as follows: First, the assembly process of the entire body-in-white is divided into several sequentially executed assembly stations according to its technological process, such as the chassis assembly station, the body welding station, the roof panel assembly station, etc. Inside each assembly station, the specific assembly task units that need to be cooperatively completed by multiple six-degree-of-freedom robotic arms are further analyzed in detail. For example, in the body welding station, the welding task of the left side panel and the floor panel is analyzed, and it is determined that two robotic arms are required to separately grasp the components and cooperate with the welding torch to complete the welding operation. Analyze the specific motion sequences of each robotic arm in each assembly task unit, including the starting posture, motion trajectory, target posture of the robotic arm, the actions of the end effector (such as the welding parameter settings of the welding torch, the clamping and loosening of the fixture), as well as the cooperative motion time sequence with other robotic arms. By analyzing the component connection relationships and assembly process requirements in the assembly schematic diagram, determine the cooperation paths of multiple robotic arms in three-dimensional space, and plan the safe areas and waiting areas to avoid collisions between robotic arms and between robotic arms and tooling fixtures. The analysis results are recorded in the form of a flow chart, clearly showing the start time, duration, the numbers of the participating robotic arms, and the synchronous cooperation relationships between robotic arms for each assembly task unit, and finally obtaining the multi-robot cooperation process for component assembly. For example, the flow chart shows that when welding the left side panel and the floor panel, the No. 1 robotic arm starts to clamp the left side panel at the 10th second, the No. 2 robotic arm starts to clamp the floor panel at the 11th second, and the two robotic arms simultaneously move to the welding starting point for welding operations at the 15th second. Based on the multi-robot cooperation process for component assembly obtained in step S12, acoustic emission sensors are rigidly fixed and installed on the end flanges of each six-degree-of-freedom robotic arm participating in the body structure assembly through high-strength bolts. This sensor has the ability to receive acoustic emission signals with a wide frequency band, and its frequency response range covers 10 kHz to 1 MHz, and the sensitivity is -65 dBV / μbar. During the process of multi-robot cooperation for synchronous assembly of the body structure, start all the participating robotic arms and execute operations according to the predetermined motion trajectories and assembly actions, such as performing spot welding, arc welding, or bolt tightening, etc.Meanwhile, through the signal conditioner and data acquisition system connected to the sensors, the structural vibration acoustic emission signals generated by each robotic arm during the assembly task are synchronously collected at a sampling frequency of 500 kHz. The acquisition duration covers the entire multi-robot collaborative assembly process. For each acquired acoustic emission time-domain signal, spectral analysis is performed using the Fast Fourier Transform (FFT) algorithm to convert the time-domain signal to the frequency domain, obtaining the structural vibration frequency components and their amplitudes corresponding to different time points. During the multi-robot collaborative process, the working intervals of multiple six-degree-of-freedom arms are not long. The vehicle body vibration caused by the previous robotic arm during its operation will affect the assembly control of the six-degree-of-freedom arm that is about to perform assembly. For example, within 0.1 seconds after welding is completed at a certain welding point, the acquired acoustic emission signal is analyzed by FFT, and the main structural vibration frequency components during this period are concentrated around 2 kHz, 5 kHz, and 8 kHz, with corresponding amplitudes of 0.5 mV, 0.8 mV, and 0.3 mV respectively. The vibration frequency data collected from all robotic arms participating in the assembly are integrated to obtain a set containing the structural vibration frequency information of multiple robotic arms at different time points during the entire assembly process, that is, the structural vibration frequency of the multi-robot assembled vehicle body. Time-series tags are inserted into the structural vibration frequency data of the multi-robot assembled vehicle body obtained in step S13. The specific operation is to accurately timestamp the structural vibration frequency data collected at each time point with the specific assembly task unit being executed at that time and the time point of this task unit according to the multi-robot collaborative process of component assembly recorded in step S12. For example, when the 5th solder joint welding operation is performed on the chassis and the left side panel, the collected vibration frequency data will be labeled with the tag "chassis - left side panel welding - solder joint 5 - 5.2 s", where "5.2 s" indicates that this data was collected at the 5.2nd second after the start of the entire assembly process. For the vibration frequency data collected by different robotic arms, the corresponding robotic arm numbers also need to be labeled. For example, the data collected by robotic arm 1 will be labeled with "robotic arm 1". In this way, the structural vibration frequency data is accurately associated with the specific events during the assembly process, forming the assembly structure time-series vibration frequency with time-series and assembly event information. Missing value filling is performed on the assembly structure time-series vibration frequency data obtained in step S14. During the actual acoustic emission signal acquisition process, due to reasons such as electromagnetic interference, instantaneous sensor failures, or data transmission interruptions, the vibration frequency data at some time points may be missing. For these missing data points, linear interpolation is used for filling.

[0026] In another embodiment, during the process of obtaining the assembly schematic diagram of automotive parts, a high-resolution industrial two-dimensional scanning system is used to perform standardized digital acquisition on the vehicle structure engineering drawings. The body side panel assembly, front longitudinal beam, mounting bracket, and engine mounting bolt are selected as the main connection structures. The schematic diagram data sources include the CAD structure diagram and the standard assembly instruction manual provided by the vehicle factory, which are uniformly converted into a vector image format containing structure numbers, connection position coordinates, bolt models, screwing directions, pre-tightening force reference values, and assembly sequences. Each schematic diagram is decomposed into a structural hierarchical node diagram, with the attributes of the connecting parts, spatial topological relationships, and torque requirement distributions marked, and layer identifiers are added for subsequent multi-station collaborative process mapping. The standardized resolution of the image data is 300 dpi, and the format is uniformly an SVG vector diagram structure that can be embedded in a flowchart parser.

[0027] Perform a multi-robot collaboration process analysis on the obtained assembly schematic diagram of automotive parts. Use the static process flow coding method to deconstruct the process nodes of the timing sequence of threaded connectors in the six-degree-of-freedom robotic arm assembly process, and use the multi-station collaborative mapping method to splice the assembly station numbers, robotic arm numbers, action time periods, and idle waiting times corresponding to each assembly bolt. During the specific operation process, the screwing task is divided into five stages: bolt picking, alignment, initial tightening, positioning, and final tightening. Each stage is associated with one or more robotic arms to execute subtasks, forming an assembly timing table. For example, in the front mounting bracket area, robotic arm A1 is responsible for performing the initial tightening, and robotic arm A2 performs the final tightening. The time overlap period is 0.4 seconds, indicating that there is a cross-action window within 0.4 seconds in this assembly process, which belongs to the collaborative control section. On this basis, establish a robotic arm assembly action queue diagram. The output multi-robot collaboration process of the parts assembly includes the assembly sequence of each connecting part, the collaborative station relationship diagram, the task execution time window, and the operation priority index.

[0028] Based on the obtained multi-robot collaboration process, the acoustic emission sensor integrated at the tool flange at the end of the six-degree-of-freedom robotic arm is used to collect the real-time response of the vehicle body structure during the assembly process, and obtain acoustic emission characteristics such as micro-crack sound waves, metal slip sound signals, material contact friction waves, and mechanical shock waves generated during the assembly connection process. The sensor model is a piezoelectric wide-band acoustic emission probe, with a working frequency range of 50 kHz to 800 kHz and a sampling frequency of 2 million points per second. During the entire process of the robotic arm performing the screw-tightening operation, the acoustic emission sensor continuously records the vibration signals during the assembly process, especially focusing on the vibration response changes of different structural components in the initial tightening - transition - final tightening stages. To ensure the structural relevance of the data, after each screwing is completed, the system immediately records the current structure number, the six-axis coordinates of the robotic arm pose, the bolt model used, and the corresponding acoustic emission time series segment. The finally output data is the vibration frequency data of the multi-robot assembled vehicle body structure, including characteristic indicators such as the vibration frequency response spectrum line, acoustic wave peak frequency, waveform rising edge slope, and vibration duration of each connecting piece within a specific time period. During the multi-robot collaboration process, the working time intervals between multiple six-degree-of-freedom arms are not long, and the vehicle body vibration caused by the previous robotic arm during operation will affect the assembly control of the six-degree-of-freedom arm to be assembled.

[0029] Insert time series tags into the collected vibration frequency data of the multi-robot assembled vehicle body structure. Adopt a method based on aligning the built-in timestamp of the robotic arm control system with the recording time of the acoustic emission signal, and mark the robotic arm operation actions (such as starting screwing, thread contact, torque threshold point) as key time points on the corresponding frequency signals. During the process of inserting tags, all signal segments are segmented based on the signal intensity mutation points, and the signal segments are divided into five categories: "no-load segment", "pre-contact segment", "transition force application segment", "maximum load segment", and "torque stable segment". An action tag is inserted at the front end of each segment, and the tag format includes the connecting piece number, robotic arm number, action stage mark, relative start time, current control axis data, and frequency segment number.

[0030] For the situation of missing frequency data in the time series vibration frequency data of the assembly structure due to accidental interference of the sensor or signal interruption, a method combining interpolation completion and adjacent spectrum inference is used to fill in the missing values. First, identify the missing segments in the signal, set the threshold of the missing segment length to a short-time missing section within 20 milliseconds, and use the linear interpolation method to fill in the missing data based on the mean and change slope of the two segments before and after the missing segment. For long-time missing segments longer than 20 milliseconds, use the frequency peak trajectory and response mode in the front and rear structure segments for feature fitting, and combine the frequency spectrum characteristics of the corresponding bolts of the structural components on the connecting pieces of the same type of material for spectrum inference and completion.

[0031] Step S2 includes the following steps: Step S21: Conduct an asymmetric analysis on the vibration frequency filled in the assembly body structure time series to obtain the asymmetric dynamic frequency of the assembly body structure; Step S22: Based on the assembly schematic diagram of automotive parts and the asymmetric dynamic frequency of the assembly body structure, simulate the disorder of the vibration azimuth impact at the assembly connection structure interface to obtain the disordered data of the vibration azimuth impact force at the connection interface; Step S23: Based on the assembly schematic diagram of automotive parts and the disordered data of the vibration azimuth impact force at the connection interface, conduct a misalignment regression analysis of the connection pre-tightening force to obtain the misalignment regression data of the connection pre-tightening force; Step S24: According to the disordered data of the vibration azimuth impact force at the connection interface and the misalignment regression data of the connection pre-tightening force, predict the fatigue fracture failure of the connection structure to obtain the fatigue fracture prediction data of the connection structure.

[0032] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: Conduct an asymmetric analysis on the vibration frequency filled in the assembly body structure time series to obtain the asymmetric dynamic frequency of the assembly body structure; In the embodiment of the present invention, for the vibration frequency data filled in the assembly body structure time series caused by the assembly process of other six-degree-of-freedom robotic arms obtained in step S15, first calculate the power spectral density of the spectrum, which reflects the energy intensity of different frequency components. Then, calculate the first moment (mean), second moment (variance), third moment, and fourth moment of the spectrum. Based on these moments, calculate the skewness coefficient and kurtosis coefficient of the spectrum. The skewness coefficient reflects the skewness degree of the spectrum distribution relative to the mean. A positive skewness indicates that the distribution is skewed towards the high-frequency direction, and a negative skewness indicates that the distribution is skewed towards the low-frequency direction. The kurtosis coefficient reflects the sharpness degree of the spectrum distribution. A high kurtosis indicates that the distribution is concentrated near the mean and has a long tail, and a low kurtosis indicates that the distribution is relatively flat. The calculation formulas are: skewness = third central moment / (1.5th power of the second central moment), kurtosis = fourth central moment / (square of the second central moment). Calculate the above skewness and kurtosis coefficients for the spectrum at each time point during the tightening process to obtain the sequences of skewness and kurtosis changing with time. These two time series together constitute the asymmetric dynamic frequency of the assembly body structure, and this data specifically describes the dynamic asymmetric change of the energy distribution of the body structure vibration frequency during the bolt tightening process. For example, when the bolt starts to be screwed in, it is observed that the skewness coefficient is close to zero and the kurtosis coefficient is close to 3 (approximate normal distribution), while during the process of gradually increasing the bolt pre-tightening force, due to the changes in friction and contact state, the energy distribution of the vibration spectrum becomes asymmetric. For example, the skewness coefficient increases to 0.5 and the kurtosis coefficient increases to 4.2, indicating a significant increase in the energy of the high-frequency vibration components.

[0033] Step S22: Based on the assembly schematic diagram of automotive parts and the asymmetric dynamic frequency of the assembled body structure, simulate the randomness of the vibration azimuth impact of the assembly connection structure interface to obtain the random data of the vibration azimuth impact force of the connection interface; In the embodiment of the present invention, based on the detailed information about bolt connection in the automotive parts assembly schematic diagram obtained in step S11, such as the model specification of the bolt (M8×20), pitch (1.25 mm), and the material properties of the connecting parts (for example, the elastic modulus of steel is 200 GPa and the Poisson's ratio is 0.3), and the asymmetric dynamic frequency of the assembled body structure obtained in step S21, simulate the randomness of the vibration azimuth impact of the assembly connection structure interface. The specific operation is to simplify the contact interface between the bolt and the connecting parts into multiple tiny contact units. For each time point, according to the skewness and kurtosis values of the asymmetric dynamic frequency corresponding to this time point, assume that the direction of the instantaneous impact force generated on each contact unit is a vector randomly distributed in three-dimensional space. The statistical characteristics of this random distribution (such as the mean direction, standard deviation of the direction) are associated with the skewness and kurtosis values of the asymmetric dynamic frequency. For example, a higher skewness value causes the impact force direction to be more concentrated within a certain specific angular range, while a higher kurtosis value causes the distribution of the impact force direction to be more sharp. Through the Monte Carlo simulation method, generate a large number of impact force direction samples that conform to this random distribution. Assume that the magnitude of the impact force received by each contact unit is proportional to the total vibration energy at this time point. Then, perform vector superposition on the impact forces with random directions generated on each contact unit to obtain the resultant impact force received by the entire bolt connection interface at a specific time point. Since the direction of the impact force on each contact unit is random, the direction of the resultant impact force received by the entire connection interface will also show randomness over time and different contact units. To quantify this randomness, calculate the eigenvalues of the direction cosine matrix of the impact force direction vectors on all contact units at each time point. The magnitude of the eigenvalue reflects the degree of dispersion of the impact force direction in each main direction. If the eigenvalue distribution is relatively uniform, it indicates a higher randomness of the impact force direction; if there is one or several significantly larger eigenvalues, it indicates an obvious dominant direction of the impact force direction and lower randomness. Through the above simulation and calculation for the entire bolt tightening process, obtain the random data of the vibration azimuth impact force of the connection interface, which includes the magnitude and direction of the resultant impact force on the bolt connection interface at different time points, as well as the quantified randomness index (for example, the minimum eigenvalue of the direction cosine matrix). For example, when the bolt just touches the surface of the connecting part, due to unstable contact, the randomness index of the impact force direction is relatively high (for example, the minimum eigenvalue is close to 0). As the tightening process progresses, the contact gradually stabilizes, the impact force direction gradually tends to be consistent, and the randomness index decreases (for example, the minimum eigenvalue increases to 0.6).

[0034] Step S23: Based on the assembly schematic diagram of automotive parts and the disordered data of the vibration azimuth impact force at the connection interface, perform a misalignment regression analysis of the connection pre-tightening force to obtain misalignment regression data of the connection pre-tightening force; In the embodiment of the present invention, based on the theoretical pre-tightening torque parameter (for example, 40 Nm) of the bolt connection in the assembly schematic diagram of automotive parts obtained in step S11 and the disordered data of the vibration azimuth impact force at the connection interface obtained in step S22, a misalignment regression analysis of the connection pre-tightening force is performed. First, the theoretical pre-tightening torque is converted into the theoretical axial pre-tightening force. The conversion formula is: , where F is the axial pre-tightening force, T is the pre-tightening torque, d is the nominal diameter of the bolt, and K is the torque coefficient, the value of which depends on the friction coefficient of the bolt and the thread parameters. For example, for a galvanized steel bolt, the torque coefficient is usually between 0.15 and 0.2, and 0.18 is taken. Then, analyze the disordered data of the vibration azimuth impact force at the connection interface obtained in step S22, and extract the key features characterizing the impact disorder, such as the minimum eigenvalue of the impact force direction cosine matrix, the mean and variance of the resultant impact force, etc. Assume that the loss of the bolt pre-tightening force is related to the disorder degree of the vibration impact received by the connection interface and the magnitude of the impact force. Establish a multiple linear regression model, with the extracted impact disorder features and impact force statistical features as independent variables and the misalignment amount of the pre-tightening force (the difference between the actual pre-tightening force and the theoretical pre-tightening force) as the dependent variable. Determine the coefficients of the regression model by training the model with a large amount of experimental data or finite element simulation data. The training data is obtained from bolt tightening experiments under different vibration conditions and measuring the actual pre-tightening force loss amount. For example, the experimental data includes the pre-tightening force loss amount of M8 bolts under different vibration frequencies, amplitudes, and disordered impact effects. Input the impact disorder feature data of the specific bolt connection interface obtained in step S22 into the trained regression model to predict the misalignment amount of the pre-tightening force that occurs in this bolt connection under the current assembly conditions, and obtain the misalignment regression data of the connection pre-tightening force. For example, the regression model predicts that under the current vibration impact disorder condition, the pre-tightening force of this M8 bolt will be lost by 5 Nm.

[0035] Step S24: Based on the disordered data of the vibration azimuth impact force at the connection interface and the misalignment regression data of the connection pre-tightening force, perform a prediction of the fatigue fracture failure of the connection structure to obtain fatigue fracture prediction data of the connection structure.

[0036] In the embodiments of the present invention, based on the disordered data of the vibration azimuth impact force of the connection interface obtained in step S22 and the misaligned regression data of the connection pre-tightening force obtained in step S23, a prediction of the fatigue fracture failure of the connection structure is carried out, specifically for the fatigue failure of bolt connections. First, determine the fatigue strength limit and S-N curve parameters of the bolt material. These parameters can usually be obtained from material manuals or experimental data. For example, the fatigue strength limit of a certain type of steel bolt is 200 MPa at 10^6 cycles, and the slope of the S-N curve is -0.2. Then, based on the disordered data of the vibration azimuth impact force of the connection interface obtained in step S22, analyze the dynamic stress suffered by the bolt during the tightening process. This includes the mean stress generated by the pre-tightening force and the alternating stress caused by vibration impact. The magnitude of the alternating stress is proportional to the magnitude of the impact force, and its direction and distribution are related to the direction and disorder of the impact force. For example, if the direction of the impact force is highly disordered, different parts of the bolt will bear alternating stresses in different directions and magnitudes. Combining the misaligned regression data of the pre-tightening force obtained in step S23, correct the actual pre-tightening force of the bolt connection and recalculate the mean stress generated by the pre-tightening force. The actual mean stress will decrease due to the loss of the pre-tightening force. Then, use the Miner linear cumulative damage theory to predict the fatigue life of the bolt. Decompose the complex time-varying stress spectrum suffered by the bolt during the tightening process into several cyclic loads with different stress amplitudes. For each stress amplitude, determine its fatigue life (number of cycles) according to the S-N curve of the bolt. Then, calculate the fatigue damage caused by each stress level based on the actual number of stress cycles and the corresponding fatigue life. Linearly accumulate the fatigue damage caused by all stress levels to obtain the total fatigue cumulative damage. When the cumulative damage value reaches 1, it is predicted that the bolt connection will suffer fatigue fracture failure. For example, it is predicted that after a certain bolt completes the tightening operation, due to the alternating stress caused by the loss of the pre-tightening force and vibration impact, its fatigue cumulative damage will reach the fatigue limit after 10^5 cycles, thus predicting its fatigue life to be 100,000 cycles. Obtain the fatigue fracture prediction data of the connection structure, which includes the possibility of fatigue fracture or the predicted fatigue life of each bolt connection under the current assembly conditions.

[0037] Step S22 includes the following steps: Step S221: Extract the morphological structure of the connecting piece from the assembly schematic diagram of the automotive parts to obtain the morphological structure data of the connecting piece; Step S222: Perform a coupling analysis of the vibration transfer of the connecting piece on the morphological structure data of the connecting piece according to the asymmetric dynamic frequency of the assembled body structure to obtain the vibration transfer coupling data of the connecting piece; Step S223: Fit the random process of the vibration azimuth of the vibration transfer coupling data of the connecting piece to obtain the random process data of the vibration azimuth of the connecting piece; Step S224: Based on the random process data of the vibration orientation of the connecting piece, the vibration transfer coupling data of the connecting piece, and the asymmetric dynamic frequency of the assembled body structure, perform superposition fitting of the impact force in the connection interface orientation to obtain the superposition data of the impact force in the connection interface orientation. Step S225: According to the superposition data of the impact force in the connection interface orientation, perform simulation on the disorder of the vibration orientation impact of the assembled connection structure interface to obtain the disorder data of the vibration orientation impact of the connection interface.

[0038] In the embodiment of the present invention, all the parts connected by bolts in the obtained assembly schematic diagram of automotive parts are identified and data is extracted. For each bolt connection, its detailed morphological structure parameters are extracted, including the model specification of the bolt (for example, M8×20, indicating a bolt with a nominal diameter of 8 mm and a length of 20 mm), the type of thread (for example, fine thread, coarse thread), the pitch (for example, 1.25 mm), the head shape (for example, hexagon head, internal hexagon head), the use of washers (for example, whether to use flat washers, spring washers, and the size specifications of the washers), as well as the material properties of the connecting piece (for example, the grade of steel, elastic modulus of 200 GPa, Poisson's ratio of 0.3), the roughness of the connection surface, and other information.

[0039] Based on the skewness and kurtosis data of the assembly body structure's asymmetric dynamic frequency caused by the assembly process of the other six-degree-of-freedom robotic arm obtained in step S21, a coupling analysis of the vibration transmission of the connecting parts is carried out on the morphological structure data of each bolt connection extracted in step S221. The specific operation is as follows: for each bolt connection, a finite element model is constructed. This model accurately simulates the geometric shapes and material properties of the bolt, washer (if any), and the connected parts. The skewness and kurtosis data obtained in step S21 are applied as excitation loads to the finite element model. The specific application method is to associate the skewness and kurtosis values with the frequency components and energy distribution of the vibration signal, simulating the asymmetric vibration generated at the contact interface of the connecting parts due to the vibration of the assembly body structure and the interaction between the connecting parts during the bolt tightening process. Analyze the transmission efficiency and attenuation characteristics of the vibration energy at different parts of the bolt (such as the bolt head, thread, screw rod, connection interface) and the connected parts under different times and different asymmetric vibration excitations. Calculate the transmission coefficients of the vibration in different directions (such as axial, radial, torsional directions), as well as the change in the vibration frequency. For example, analyze that for a certain M8 bolt connection under a vibration excitation with a skewness of 0.6 and a kurtosis of 4.0, the vibration amplitude transmitted from the connected part to the bolt head is attenuated by 20%, and the attenuation of the high-frequency components is less than that of the low-frequency components. For each bolt connection, the time-series data of the transmission efficiency of the vibration with different frequency components in different directions during the entire tightening process are obtained, that is, the coupling data of the vibration transmission of the connecting parts. This data includes the amplification or attenuation characteristics of each bolt connection for vibrations with different frequencies at different time points, as well as the coupling relationship of the vibration directions.

[0040] Perform a vibration azimuth random process fitting on the vibration transfer coupling data of each bolt-connected component obtained in step S222. The specific operation is as follows: for each contact interface of the bolt connection, assume that the vibration direction of each tiny area on it is a random variable. Based on the vibration transfer efficiency in different directions obtained in step S222, determine the probability distribution parameters of this random variable. For example, if the analysis shows that at a certain time point, the vibration transfer efficiency in the axial direction is much higher than that in the radial and torsional directions, then assume that the vibration direction of the tiny area of this contact interface is more inclined to the axial direction. Using statistical methods, such as maximum likelihood estimation, according to the proportional relationship of the vibration amplitude in different directions in the vibration transfer coupling data, fit the probability distribution model of the vibration direction of each tiny area of the contact interface. For example, the Von Mises distribution is used to describe the probability distribution of directions on a circle or a sphere. This model can describe the mean direction and the degree of dispersion of the vibration direction. For each bolt connection, at each time point during the entire tightening process, a corresponding vibration azimuth random process model is fitted to obtain the vibration azimuth random process data of the component. This data contains the parameters that describe the probability distribution model of the vibration direction of each tiny area of the bolt connection contact interface. These parameters change with time and are affected by the asymmetric dynamic frequency. For example, at the initial stage of tightening, due to unstable contact, the probability distribution of the vibration direction is relatively uniform. As the pre-tightening force increases, the contact gradually stabilizes, and the probability distribution of the vibration direction becomes more concentrated. Based on the vibration azimuth random process data of each bolt-connected component obtained in step S223, the vibration transfer coupling data of the component obtained in step S222, and the asymmetric dynamic frequency of the assembled body structure obtained in step S21, perform a superposition fitting of the impact force in the connection interface azimuth. The specific operation is as follows: for each contact interface of the bolt connection, discretize it into a large number of tiny units. At each time point, according to the vibration azimuth random process model obtained in step S223, randomly generate a vibration direction vector for each tiny unit. This direction vector follows the probability distribution described by this model. Then, according to the vibration transfer coupling data obtained in step S222, determine the vibration amplitude on each tiny unit. Assume that the magnitude of the impact force is proportional to the vibration amplitude, and the direction of the impact force is the same as the vibration direction. Superimpose the impact force vectors on all tiny units to obtain the resultant impact force vector received by the entire bolt connection interface at this time point. Since the vibration direction of each tiny unit is randomly generated, the result obtained by each superposition will be different. In order to obtain a statistically significant resultant impact force, multiple Monte Carlo simulations need to be performed, that is, repeat the above process of random generation and superposition multiple times (for example, 1000 times). Perform a statistical analysis on the resultant impact force vector obtained by each simulation, and calculate its mean value and covariance matrix. The mean value reflects the main direction of the resultant impact force, and the covariance matrix reflects the degree of dispersion of the resultant impact force direction.In addition, the initial impact force magnitude generated by each micro-unit can be weighted according to the skewness and kurtosis values of the asymmetric dynamic frequency obtained in step S21 to reflect the influence of asymmetric vibration on the impact force magnitude. By performing the above operations at each time point of the entire bolt tightening process, the superimposed fitting data of the impact force in the connection interface orientation is obtained. This data includes the mean vector and covariance matrix of the combined impact force received by each bolt connection interface at different time points. According to the superimposed data of the impact force in the connection interface orientation of each bolt connection obtained in step S224, a simulation of the disorder of the impact in the vibration orientation of the assembly connection structure interface is carried out. The specific operation is that for each bolt connection, at each time point, the mean vector and covariance matrix of the combined impact force obtained in step S224 are analyzed. The impact disorder can be quantified by analyzing the distribution divergence of the combined impact force direction. One method is to calculate the variance of the projection of the unit vector of the combined impact force direction in each direction. If the variance is large, it indicates that the direction of the combined impact force is relatively dispersed and the disorder is high; if the variance is small, it indicates that the direction of the combined impact force is relatively concentrated and the disorder is low. Another method is to calculate the trace of the covariance matrix of the combined impact force. The larger the trace, the greater the fluctuation of the combined impact force in each direction and the higher the disorder. The entropy value of the combined impact force direction can also be calculated. The larger the entropy value, the more uniform the direction distribution and the higher the disorder. For each bolt connection, during the entire tightening process, the disorder index of the combined impact force direction at its connection interface is calculated to obtain the disorder data of the impact in the vibration orientation of the connection interface. This data includes the quantification index of the disorder of the impact in the vibration orientation of each bolt connection at different time points. For example, when the bolt starts to be tightened, due to unstable contact, the impact force direction is relatively chaotic and the disorder index is high (for example, the trace of the direction cosine matrix is large). As the pre-tightening force increases, the contact gradually stabilizes, the impact force direction tends to be consistent, and the disorder index decreases (for example, the trace of the direction cosine matrix decreases). By analyzing the change of this disorder index over time, the regularity and randomness of the vibration impact direction received by the connection interface during the bolt tightening process can be understood.

[0041] Step S224 includes the following steps: Calculate the frequency response acceleration ratio of the vibration transfer coupling data of the connecting piece to obtain the vibration frequency acceleration ratio of the connecting piece; Based on the random process data of the vibration orientation of the connecting piece, perform multi-orientation frequency diffusion effect coupling processing on the vibration frequency acceleration ratio of the connecting piece to obtain the coupling data of the orientation frequency diffusion effect; Perform frequency vibration energy intensity diffusion radius interpolation processing on the coupling data of the orientation frequency diffusion effect to obtain the interpolation data of the frequency intensity diffusion radius; Based on the interpolation data of the frequency intensity diffusion radius and the asymmetric dynamic frequency of the assembled body structure, the equal gradient of the spatial energy density vector between the connection interfaces is identified to obtain the equal gradient of the spatial energy density vector. Based on the equal gradient of the spatial energy density vector, the superposition fitting of the azimuth impact force on the connection interface is carried out to obtain the superposition data of the azimuth impact force on the connection interface.

[0042] In the embodiment of the present invention, first, the frequency response acceleration ratio of the vibration transfer coupling data of the connecting member obtained in step S222 is calculated. The specific operation is that for each bolt connection, the acceleration responses of different parts of the connecting member (for example, the contact surface of the connected member, the bolt head) under vibration excitation at different frequencies are analyzed. The ratio of the acceleration amplitude of a certain point on the contact surface of the connected member to the acceleration amplitude of the corresponding point on the bolt head at each frequency point is calculated to obtain the frequency response acceleration ratio. This ratio reflects the transmission efficiency of vibration in the connecting member and the amplification or attenuation of different frequency components of vibration during transmission. For example, at 100 Hz, the acceleration of the contact surface of the connected member is and the acceleration of the bolt head is , then the acceleration ratio at this frequency is 4. The above calculation is performed for the entire frequency range (for example, 10 Hz to 10 kHz) to obtain a curve of the vibration frequency acceleration ratio of the connecting member varying with frequency. This curve describes the acceleration transmission characteristics of vibration signals at different frequencies passing through the bolt connection. Then, based on the vibration azimuth random process data obtained in step S223, the above-mentioned vibration frequency acceleration ratio of the connecting member is subjected to multi-azimuth frequency diffusion effect coupling processing. The specific operation is that for the contact interface of each bolt connection, the randomness of the vibration directions in different micro-regions thereon is considered. According to the vibration azimuth probability distribution model fitted in step S223, the diffusion effect occurring in different directions when the vibration energy is transmitted from the connected member to the bolt at each frequency point is determined. For example, if the transmission efficiency of the vibration at a certain frequency is higher in the axial direction and lower in the radial direction, and the randomness of the vibration direction is larger, then after the vibration energy of this frequency is transmitted to the bolt, its direction distribution will be more dispersed. Using convolution operation, the frequency response acceleration ratio is coupled with the statistical characteristics of the vibration azimuth random process (for example, the variance of the direction distribution). The coupling result obtains the azimuth frequency diffusion effect coupling data, which describes the change in the direction distribution of the vibration acceleration after passing through the bolt connection at different frequencies. For example, the standard deviation of the acceleration direction of the vibration at a certain frequency increases from 5 degrees to 15 degrees after passing through the bolt connection.

[0043] Next, interpolation processing is performed on the azimuth frequency diffusion effect coupling data for the frequency vibration energy intensity diffusion radius. The specific operation is that for each contact interface of the bolt connection, the diffusion range of the vibration energy intensity at different frequencies in space is represented by an equivalent diffusion radius. The size of this diffusion radius is related to the acceleration amplitude at this frequency and the divergence of the direction distribution. According to the azimuth frequency diffusion effect coupling data, using an interpolation algorithm (e.g., linear interpolation, spline interpolation), estimate the diffusion radius of the vibration energy intensity on the connection interface over the entire frequency range. For example, in the low-frequency band, the vibration energy is concentrated near the contact surface and the diffusion radius is small; in the high-frequency band, due to effects such as structural resonance, the vibration energy diffuses over a larger range and the diffusion radius is large. The interpolation processing obtains the interpolation data of the frequency intensity diffusion radius, which describes the spatial diffusion range of the vibration energy at different frequencies on the connection interface.

[0044] Then, based on the above interpolation data of the frequency intensity diffusion radius and the asymmetric dynamic frequency of the assembled body structure obtained in step S21, identify the equal gradient of the spatial energy density vector between the connection interfaces. The specific operation is that for each bolt connection, at each time point, analyze its influence on the spatial distribution of the vibration energy according to its corresponding asymmetric dynamic frequency (skewness and kurtosis). For example, a higher skewness causes the energy distribution to be more concentrated in a certain direction, while a higher kurtosis causes an increase in the sharpness of the energy distribution. Combining the interpolation data of the frequency intensity diffusion radius, identify the equal gradient distribution of the spatial energy density vector between the connection interfaces. The magnitude and direction of the gradient reflect the direction of energy flow and the rate of change of intensity. For example, in the thread contact area of the bolt, due to the action of the pre-tightening force, the energy density gradient is large.

[0045] Finally, based on the identified equal gradient of the spatial energy density vector, perform superposition fitting of the azimuth impact force on the connection interface. The specific operation is to discretize the connection interface into multiple tiny units. For each tiny unit, the magnitude of the impact force it receives is proportional to the magnitude of the spatial energy density gradient at this unit, and the direction of the impact force is the same as the direction of the energy density gradient. Superimpose the impact force vectors on all the tiny units to obtain the resultant impact force vector received by the entire bolt connection interface at this time point. Since the distribution of the spatial energy density gradient is affected by the asymmetric dynamic frequency and the vibration diffusion effect, the resultant impact force obtained by superposition also reflects the action of these factors. By performing the above operations at each time point of the entire bolt tightening process, obtain the superposition data of the azimuth impact force on the connection interface, which contains the resultant impact force vectors received by each bolt connection interface at different time points.

[0046] Step S23 includes the following steps: Step S231: Extract the theoretical assembly connection torque pre-tightening force based on the assembly schematic diagram of automotive parts to obtain the theoretical assembly connection torque pre-tightening force; Step S232: Conduct multi-scale analysis of the azimuth vibration impact force on the disordered data of the connection interface vibration azimuth impact force to obtain the multi-scale data of the azimuth vibration disorder impact force; Step S233: Analyze the spatial dimension impact force fluctuation skewness difference of the multi-scale data of the azimuth vibration disorder impact force to obtain the spatial impact force fluctuation skewness difference data; Step S234: Perform pre-tightening force interference misalignment gradient simulation identification on the theoretical assembly connection torque pre-tightening force according to the spatial impact force fluctuation skewness difference data to obtain the pre-tightening force interference misalignment gradient data; Step S235: Conduct connection pre-tightening force misalignment regression analysis on the pre-tightening force interference misalignment gradient data to obtain the connection pre-tightening force misalignment regression data.

[0047] In the embodiments of the present invention, for the assembly schematic diagram of automotive parts, for all assembly parts connected by bolts, the theoretical assembly connection torque pre-tightening force values specified in their design specifications are extracted. These values are usually directly marked in the part list, assembly process description or relevant technical documents of the assembly schematic diagram. For example, for the M12×35 specification bolts connecting the chassis and the subframe, its theoretical pre-tightening torque is specified as 80 N·m; for the M8×20 specification bolts fixing the door hinge, its theoretical pre-tightening torque is 25 N·m. These theoretical pre-tightening torque values are determined through design calculations based on factors such as the strength requirements of the connection, the material and specification of the bolts, and the force conditions of the connecting parts, aiming to ensure the reliability and tightness of the connection. The extraction operation needs to traverse each bolt connection in the assembly schematic diagram and read the relevant theoretical pre-tightening torque parameters. Perform multi-scale analysis of the azimuth vibration impact force on the disordered data of the vibration azimuth impact force at the connection interface. First, expand the impact force vector data of each bolt node in terms of direction, divide the impact force direction interval in units of every 10 degrees according to the spherical coordinate system, and a total of 648 direction sub-intervals are obtained. For each direction, the duration, peak acceleration and impact frequency of the impact signal are statistically analyzed, and multi-scale wavelet decomposition processing is performed on the three parameters respectively. The Daubechies 4th-order wavelet is used for four-layer decomposition to extract the impact energy density in each frequency band. On this basis, through direction accumulation processing, the impact force energy distribution curve at each direction scale is obtained, and after normalization, the multi-scale data of the azimuth vibration disordered impact force is formed. This data structure includes: bolt number, impact direction angle (azimuth angle and pitch angle), multi-scale energy density vector, frequency band index and signal change trend label. Perform spatial dimension impact force fluctuation skewness difference analysis on the multi-scale data of the azimuth vibration disordered impact force. The specific operation is to consider the differences in the impact forces received by different regions on the connection interface for each bolt connection.

[0048] Perform spatial dimension impact force fluctuation skewness difference analysis on azimuth vibration disorder impact force multi-scale data. During the operation process, first, perform three-dimensional reconstruction on the multi-scale impact energy data of all bolt nodes, form a three-dimensional tensor structure with the direction angle, frequency scale, and energy density, and perform skewness extraction operation based on local extreme value analysis on this structure. The specific method is as follows: Taking each frequency scale as a layer, traverse all direction sub-intervals, and statistically analyze the energy peak azimuth of the direction distribution and the angle difference between it and the theoretical axial direction. By comparing the variances of the energy concentration directions and the skewness direction distribution densities at different frequency scales, quantify the direction fluctuation inconsistencies between scales. Further use the entropy difference analysis method for these skewness distribution characteristics to extract the direction perturbation amplitude of the spatial impact force fluctuation, and consider the perturbation direction angle deviation greater than 25 degrees as a significant skewness. Finally, generate spatial impact force fluctuation skewness difference data, and the fields include: bolt number, scale number, maximum perturbation direction angle, perturbation amplitude, spatial skewness index, and skewness change trend.

[0049] Based on the spatial impact force fluctuation skewness difference data obtained in step S233, a pre-tightening force interference misalignment gradient simulation identification is carried out on the theoretical assembly connection torque pre-tightening force extracted in step S231. The specific operation is as follows: for each bolt connection, a mechanical model is established, which takes into account the elastic deformation of the bolt, the stiffness of the connecting parts, and the friction characteristics between the connection interfaces. The impact force fluctuation amplitude and skewness difference in different regions obtained in step S233 are applied to this mechanical model as external interference loads. For example, a periodic load with a specific fluctuation amplitude and skewness distribution is applied to the thread contact area of the bolt to simulate the influence of vibration shock. Through numerical simulation methods (such as finite element analysis), the change of the pre-tightening force of the bolt connection over time and its distribution on the connection interface are simulated under the action of these interference loads. Identify the region with the largest pre-tightening force loss and the gradient change of the pre-tightening force distribution. The simulation results show that due to the non-uniform distribution of the impact force, the pre-tightening force of a certain section of the bolt thread decreases by 15%, while the change of the pre-tightening force in other parts is relatively small, forming a pre-tightening force gradient. Quantify the magnitude of the pre-tightening force loss and the non-uniformity of the distribution to obtain the pre-tightening force interference misalignment gradient data. This data includes the average loss of the pre-tightening force of each bolt connection under the action of vibration shock and the gradient distribution information of the pre-tightening force on the connection interface. A connection pre-tightening force misalignment regression analysis is carried out on the pre-tightening force interference misalignment gradient data obtained in step S234. The specific operation is as follows: for each bolt connection, a regression model is established, with the average pre-tightening force loss and the pre-tightening force gradient distribution characteristics obtained in step S234 as input variables, and the pre-tightening force misalignment amount measured actually or obtained through high-precision finite element simulation as the output variable. The type of regression model can be selected as multiple linear regression, support vector regression, or neural network, etc. Use a large amount of experimental data or high-precision finite element simulation data to train this regression model to determine the parameters of the model. The training data should cover the pre-tightening force misalignment conditions under different bolt specifications, connecting part materials, and different vibration shock conditions. After training, input the pre-tightening force interference misalignment gradient data of a specific bolt connection obtained in step S234 into the trained regression model to predict the pre-tightening force misalignment amount that occurs under the current assembly conditions of this bolt connection. Obtain the connection pre-tightening force misalignment regression data, which includes the predicted loss value of the pre-tightening force of each bolt connection. For example, it is predicted that the pre-tightening force of an M10 bolt will lose 7 N·m.

[0050] Step S234 includes the following steps: Perform an equal ratio calculation of the spatial azimuth skewness kurtosis difference on the spatial impact force fluctuation skewness difference data to obtain the azimuth skewness kurtosis difference equal ratio data; Based on the azimuth skewness kurtosis difference equal ratio data, conduct a torque pre-tightening friction force slip amount simulation evaluation on the theoretical assembly connection torque pre-tightening force to obtain the pre-tightening friction force slip amount data; Analyze the local pre-tightening force modal mismatch distribution based on the pre-tightening friction slip amount data and the ratio data such as azimuth skewness kurtosis difference, and obtain the local pre-tightening force mismatch distribution data; Based on the local pre-tightening force mismatch distribution data, perform pre-tightening force interference misalignment gradient simulation identification to obtain pre-tightening force interference misalignment gradient data.

[0051] In the embodiments of the present invention, first, based on the spatial impact force fluctuation skewness difference data obtained in the previous steps, the spatial azimuth skewness kurtosis difference ratio calculation is performed. The specific operations include: taking each bolt connection point as a unit, extracting the multi-scale impact force energy distribution data in the three-dimensional spherical direction, and dividing the direction angle into two dimensions: horizontal (0° - 360°) and pitch (-90° to +90°). Using a division granularity with an angle interval of 10 degrees, a total of 648 direction units are formed. Calculate the impact energy density on each direction unit, and use the kurtosis analysis method to extract the fourth-order skewness concentration index for this distribution to evaluate the concentration intensity of the impact in certain directions. Then, perform the ratio calculation of the kurtosis values of all directions with the kurtosis values of the corresponding theoretical assembly directions to obtain the kurtosis difference ratio data. For example, in the bolt connection node of the car door interior trim panel, the theoretical assembly direction is a pitch angle of 90 degrees and a horizontal angle of 180 degrees, while the peak value of the actual impact force distribution appears in the direction of a pitch angle of 60 degrees and a horizontal angle of 150 degrees. The kurtosis value in this direction is 2.8, and the kurtosis of the theoretical direction is 1.4, then the kurtosis difference ratio value in this direction is 2. After performing such analysis on all directions, construct a kurtosis ratio spatial map to form the azimuth skewness kurtosis difference ratio data. Subsequently, based on the azimuth skewness kurtosis difference ratio data obtained above, perform a torque pre-tightening friction slip amount simulation evaluation on the theoretical assembly connection torque pre-tightening force. The specific method is as follows: First, establish a connection structure simulation model with real material parameters and friction coefficients in the finite element platform. The friction contact surface parameters are taken from the test data after the surface treatment of the bolt connection. For example, the friction coefficient of the M8 bolt connection is 0.15, and the connection materials are phosphated steel and anodized aluminum plate with thicknesses of 1.2 mm and 1.0 mm respectively. Then, according to the direction angle deviating from the theoretical assembly direction in the kurtosis difference ratio data, apply disturbing forces with different directions and amplitudes in the model. According to the moment balance condition between the slip force and the axial pre-tightening force, simulate the normal load distribution reconstruction process of the actual connection under the disturbance. By analyzing the slip threshold relationship between the normal pressure and the friction stress on the connection surface, extract the maximum slip displacement amount under the disturbance loading as the friction slip amount data generated by the torque pre-tightening mismatch. For each group of disturbance directions, record the slip start angle, maximum slip distance, percentage of the friction slip area, and the value of the normal stress reduction to form a pre-tightening friction slip amount data set with an accuracy of up to 0.01 mm. After completing the slip simulation evaluation, based on the obtained pre-tightening friction slip amount data and the kurtosis difference ratio data, analyze the local pre-tightening force mode mismatch distribution of the connection surface. This operation is based on the friction stress reconstruction information of the slip area, combined with the distribution position of the slip area in the total connection area, to deduce the change trend of the local normal pre-tightening force. Using the connection contact interface unit division method, divide the connection area into a 16×16 discrete unit array, and each unit records the change of its local normal stress value, and mark the area where the local stress reduction amplitude is greater than 20% as the mode mismatch area.In all bolted joints, construct a local pre-tightening force modal mismatch distribution map for the connection surface and classify it by mismatch levels, where the level is jointly determined by the stress reduction rate and the proportion of the slip area, forming local pre-tightening force mismatch distribution data, which includes: joint number, connection surface element number, mismatch level, normal stress change value, slip direction angle, and the continuity index of the modal mismatch area. Finally, based on this local pre-tightening force mismatch distribution data, simulate and identify the pre-tightening force interference misalignment gradient. This process maps the normal pre-tightening force loss value in the local mismatch area to the reduction ratio of the total pre-tightening force, and combines the consistency between the perturbation direction angle and the slip area direction to construct a misalignment response surface. Set multiple groups of perturbation angle combinations in the simulation platform and apply interference loads with different amplitudes (ranging from 20 N to 80 N), observe the changes in the axial stress response of the bolts in the connection, extract the gradient changes of the axial pre-tightening force decline curve, and define it as misalignment gradient data. Each simulation data includes the input perturbation direction angle, perturbation amplitude, local mismatch area ratio, and the change rate of the axial stress of the bolt. Finally, output these data as pre-tightening force interference misalignment gradient data.

[0052] Step S3 includes the following steps: Step S31: Perform convolution processing on the connection pre-tightening force misalignment regression data to obtain connection pre-tightening force misalignment convolution data; Step S32: Perform eigen-structure analysis on the connection structure fatigue fracture prediction data to obtain structure fatigue fracture eigen-data; Step S33: Perform torque-angle compound control of the six-degree-of-freedom robotic arm according to the connection pre-tightening force misalignment convolution data and the structure fatigue fracture eigen-data to generate torque-angle compound control data; Step S34: Perform iterative learning on the torque-angle compound control data to obtain torque-angle control iterative data.

[0053] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: Perform convolution processing on the connection pre-tightening force misalignment regression data to obtain connection pre-tightening force misalignment convolution data; In the embodiment of the present invention, convolution processing is performed on the connection pre-tightening force misalignment regression data obtained in step S23. The specific operation is as follows: for each connection that needs to be bolt-tightened, the pre-tightening force misalignment regression data is a sequence that changes over time, reflecting the predicted pre-tightening force loss at different stages during the tightening process. In order to extract the overall trend of this sequence and filter out high-frequency noise, one-dimensional convolution operation is used. A convolution kernel with a specific length and weight is designed. For example, a Gaussian kernel with a length of 5 time steps is selected, and its weight coefficients are calculated according to the Gaussian function, such as [0.05, 0.2, 0.5, 0.2, 0.05]. The convolution kernel is slid along the time series of the pre-tightening force misalignment, and within each sliding window, the weight coefficients of the convolution kernel are weighted and summed with the pre-tightening force misalignment values within the window to obtain the convolution result corresponding to this time step. The step size of the convolution operation is set to 1 time step to ensure that the length of the output sequence is basically the same as that of the input sequence. In this way, the original pre-tightening force misalignment time series is smoothed, highlighting the slow change trend of the pre-tightening force misalignment and suppressing the rapid fluctuations caused by factors such as instantaneous impacts. The connection pre-tightening force misalignment convolution data is obtained, which provides a smoothed pre-tightening force misalignment time series for each bolt connection, more clearly reflecting the overall loss trend of the pre-tightening force during the tightening process. For example, the original pre-tightening force misalignment data shows frequent up and down fluctuations within a certain time period, and after convolution with the Gaussian kernel, the fluctuations are smoothed, showing a gradually increasing pre-tightening force loss trend.

[0054] Step S32: Perform feature structure analysis on the connection structure fatigue fracture prediction data to obtain structure fatigue fracture feature data; In the embodiments of the present invention, characteristic structure analysis is performed on the fatigue fracture prediction data obtained in step S24. Specifically, for each bolt connection, the fatigue fracture prediction data includes information such as the predicted fatigue life (e.g., the number of cycles), the fatigue failure probability, and the fatigue damage accumulation rate. The purpose of performing characteristic structure analysis is to extract the key features in these data for use in subsequent control strategies. First, analyze the fatigue life distribution of all bolt connections to identify several key connections with the shortest fatigue life. Second, analyze the connections with a relatively high fatigue failure probability, which are the objects that the control strategy needs to focus on. In addition, the changing trend of the fatigue damage accumulation rate can also be analyzed to predict which connections will have a rapid increase in fatigue risk in the future. Integrate these analysis results to obtain the structural fatigue fracture characteristic data. This data may include a list of bolt connections sorted by fatigue life, marking the connections with a fatigue failure probability exceeding a certain threshold (e.g., 0.1), and the connections with a fatigue damage accumulation rate exceeding a certain threshold. For example, the results of the characteristic structure analysis show that the predicted fatigue life of two specific bolts connecting the car door hinge is the shortest, and the fatigue failure probability is the highest. At the same time, its fatigue damage accumulation rate is also relatively high, so it is marked as a key connection that needs to be focused on.

[0055] Step S33: Perform torque-angle composite control of the six-degree-of-freedom robotic arm based on the connection preload misalignment convolution data and the structural fatigue fracture characteristic data to generate torque-angle composite control data; In the embodiments of the present invention, the connection preload misalignment convolution data and the structural fatigue fracture characteristic data are used as input information to generate torque-angle composite control data for the six-degree-of-freedom robotic arm. In the specific operation process, first extract the perturbation response sequence and the maximum change interval of each connection node in the convolution data, and match its position index and spatial pose in the robotic arm structure model. Then extract the fatigue characteristic category, crack direction angle, and damage concentration position of the corresponding node, and map them to the influence weight of the structural weakness of the rotating joint. In the control strategy, set the basic torque input of each degree-of-freedom joint to a fixed value, such as 1.5 N·m, and then increase the torque offset value according to the misalignment convolution peak amplitude of the connection node. For example, if the peak amplitude is 2 times the original stress change rate, the control torque of the degree of freedom associated with this node increases by 0.3 N·m. At the same time, adjust the angle control amplitude according to the angle between the main crack direction in the fracture characteristics and the joint rotation direction. The smaller the angle, the more fragile the structure, and the lower the corresponding angular velocity is set. For example, when the angle is less than 10 degrees, the upper limit of the angular velocity is limited to 5 degrees / second. Finally, through the method of weight superposition, introduce the preload misalignment risk factor and the structural vulnerability index into the six-degree-of-freedom controller to dynamically correct the real-time control values of torque and angle, and generate torque-angle composite control data.

[0056] Step S34: Perform iterative learning on the torque-angle composite control data to obtain torque-angle control iterative data.

[0057] In the embodiment of the present invention, iterative learning is performed on the foregoing torque-angle composite control data. A memory iterative method based on an incremental learning structure is adopted, and the training period is set to be immediately executed after each complete assembly action of the robotic arm. Each iteration records information such as joint response delay, real-time value deviation of the connection torque, and cumulative data of fatigue stress under torque disturbance during the execution of the action. An error feedback vector is constructed with these data, and differential analysis is performed with the composite control data of the previous round. The time series difference method is used to extract the error direction and error amplitude. A momentum update mechanism is introduced, the iterative learning rate is set to 0.005, cumulative update is performed when the error directions are the same, and weight penalty correction is performed when the error directions change. The above update logic is independently executed for six degrees of freedom, and the maximum number of corrections is set to 20 times. If exceeded, the weight update process of the current degree of freedom is frozen. The final torque-angle control iterative data is obtained through the above process. The data structure includes information such as the historical control value sequence of each degree of freedom, the update trajectory of the error vector, the current learning step value, the cumulative number of corrections, the associated misalignment convolution response value, and the fatigue direction weight adjustment value, which is used to improve the accuracy and achieve mismatch adaptive response in the subsequent control system.

[0058] Step S33 includes the following steps: Step S331: Perform time-series torque limit matching for the connecting piece according to the connection pre-tightening force misalignment convolution data and the structural fatigue fracture characteristic data to obtain the time-series torque limit matching data for the connecting piece; Step S332: Perform perception of the screwing angle correction for the connecting piece based on the connection pre-tightening force misalignment convolution data to obtain the screwing angle correction perception data; Step S333: Perform segmented torque application matching for the screwing angle correction perception data according to the time-series torque limit matching data for the connecting piece to obtain the angle-associated torque segmented application data; Step S334: Perform torque-angle composite control for the six-degree-of-freedom robotic arm according to the angle-associated torque segmented application data to generate torque-angle composite control data.

[0059] In the embodiments of the present invention, based on the connection pre-tightening force misalignment convolution data obtained in step S31 and the structural fatigue fracture characteristic data obtained in step S32, connection time-sequence torque limit matching is performed to obtain connection time-sequence torque limit matching data. The specific operation is as follows: for each connection that needs to have its bolts tightened, first obtain its corresponding smoothed pre-tightening force misalignment time series. At the same time, extract the fatigue life prediction value and fatigue failure probability of this connection from the structural fatigue fracture characteristic data. If the predicted fatigue life of a certain connection is lower than a set safety threshold (for example, lower than 50,000 cycles), or its fatigue failure probability is higher than a set risk threshold (for example, higher than 0.2), then it is considered that this connection belongs to a high-fatigue-risk connection and more stringent torque limits need to be applied during the tightening process to reduce the maximum stress it bears and extend its fatigue life. For high-fatigue-risk connections, during the entire tightening process, the maximum torque value allowed for them needs to be dynamically adjusted according to their fatigue characteristics. For example, in the initial stage of tightening, a relatively large torque can be allowed to quickly establish a certain pre-tightening force; while in the stage close to the target tightening angle, the growth rate of the torque needs to be restricted to avoid generating excessive impact loads. The specific torque limit strategy can be determined based on a pre-established relationship model between fatigue life and torque peak value. For the time period with a relatively large predicted pre-tightening force loss in the pre-tightening force misalignment convolution data, the torque limit strategy also needs to be adjusted accordingly to ensure that sufficient pre-tightening force can be finally obtained. For example, in the stage with a relatively large predicted pre-tightening force loss, the allowed maximum torque value can be appropriately increased to compensate for the loss. Finally, for each bolt connection, a time-varying torque upper limit value sequence is generated. This sequence comprehensively considers the predicted trend of pre-tightening force misalignment and the fatigue risk of the connection, limits the maximum torque that the robotic arm can output during the tightening process, and obtains the connection time-sequence torque limit matching data. For example, for a connection with high fatigue risk and a relatively large predicted pre-tightening force loss, its torque limit is set to 1.2 times the theoretical value in the initial stage, and quickly drops to the theoretical value when approaching the target angle, and remains at 1.1 times the theoretical value during the time period with a relatively large predicted pre-tightening force loss.

[0060] Based on the connection pre-tightening force misalignment convolution data obtained in step S31, perform the perception of the screwing angle correction of the connecting piece to obtain the screwing angle correction perception data. The specific operation is to analyze the smoothed pre-tightening force misalignment time series for each connection that needs to have its bolts tightened. This series reflects the loss of pre-tightening force during the tightening process due to factors such as vibration and shock under ideal torque control. To compensate for this pre-tightening force loss and ensure that the desired pre-tightening force can be finally obtained, it is necessary to correct the theoretical tightening angle. The principle of correction is that if the predicted pre-tightening force loss is large, the actual tightening angle needs to be appropriately increased to obtain a higher axial pre-tightening force through greater thread deformation. The determination of the correction amount can be based on the pre-calibrated bolt torque-angle-pre-tightening force relationship curve. For example, if it is predicted that the pre-tightening force will be lost by 10% when reaching the theoretical tightening angle, a certain angle value needs to be added to the theoretical tightening angle to compensate for this part of the loss. The added angle value can be obtained by querying the calibration curve, that is, finding the angle increment that can generate an axial force equivalent to the lost pre-tightening force. In addition, the time series information of the pre-tightening force misalignment convolution data can also be used to dynamically adjust the angle correction amount. For example, in the stage where the predicted pre-tightening force loss rate is high, the rate of angle increase can be appropriately slowed down to avoid greater dynamic effects caused by too fast tightening. Finally, for each bolt connection, a time-varying sequence of screwing angle correction amounts is generated, which indicates the angle value that should be added or subtracted from the theoretical tightening angle at each tightening stage to compensate for the predicted pre-tightening force misalignment, thus obtaining the screwing angle correction perception data. For example, for a connection with an increasing predicted pre-tightening force loss, its angle correction amount will also increase, gradually increasing from an initial 0 degrees to a final 5 degrees. Based on the connection time-series torque limit matching data obtained in step S331 and the screwing angle correction perception data obtained in step S332, perform segmented torque application matching to obtain angle-related torque segmented application data. The specific operation is to divide the entire tightening process of each connection that needs to have its bolts tightened into several angle intervals. For example, the initial contact stage (0 - 30 degrees), the pre-tightening force establishment stage (30 - 180 degrees), and the final tightening stage (180 degrees to the target angle). In each angle interval, determine the target torque that the robotic arm should apply based on the torque upper limit value corresponding to this angle interval in step S331 and the angle correction amount corresponding to this angle interval in step S332. The matching principle is that at any moment, the torque output by the robotic arm cannot exceed the upper limit value specified in the time-series torque limit matching data. At the same time, the actual tightening angle needs to be adjusted based on the screwing angle correction perception data on the basis of the theoretical angle. For example, in the initial contact stage, the robotic arm can be allowed to rotate quickly with a relatively small torque until the bolt head is in full contact with the surface of the connecting piece.During the pre-tightening force establishment stage, the torque can be gradually increased, but it is necessary to ensure that it does not exceed the torque upper limit of this stage. During the final tightening stage, the target tightening angle needs to be adjusted according to the angle correction amount, and the tightening should be stopped when the corrected target angle is reached. If within a certain angle range, the torque needs to be increased according to the prediction of pre-tightening force misalignment, but the current torque is already close to or reaches the torque upper limit, it is necessary to give priority to observing the constraint of the torque upper limit to avoid damaging the connecting parts. Finally, for each bolt connection, a piecewise function relationship between the angle and the target torque is generated. This function describes the target torque values that the robotic arm should output within different tightening angle ranges, taking into account the torque upper limit and the requirements of angle correction, and obtaining the piecewise application data of angle-related torque. For example, for a certain connection, its piecewise application data of angle-related torque is defined as: within the range of 0 - 30 degrees, the target torque is 5 Nm; within the range of 30 - 180 degrees, the target torque linearly increases to 1.1 times the theoretical value, but does not exceed the set torque upper limit; within the range of 180 degrees to (theoretical angle + 5 degrees), the target torque remains at 1.1 times the theoretical value. According to the piecewise application data of angle-related torque obtained in step S333, torque-angle compound control of the six-degree-of-freedom robotic arm is carried out to generate torque-angle compound control data. The specific operation is to convert the piecewise application data of angle-related torque generated in step S333 into control instructions for the robotic arm for each connection that needs to be bolt-tightened. This instruction set includes the motion planning of each joint of the robotic arm and the torque output control of the end effector (for example, the tightening shaft) during the entire tightening process. The control instructions need to be encoded in a format that can be recognized and executed by the robotic arm control system. For example, a trajectory planning algorithm can be used to generate the spatial motion trajectory of the end effector of the robotic arm to ensure that it can smoothly approach the bolt head and apply the tightening force along the thread axis. At the same time, according to the relationship between the target torque and the angle defined in the piecewise application data of angle-related torque, the servo motor of the tightening shaft is controlled to output the corresponding torque. During the tightening process, it is necessary to monitor the angle and the actual output torque of the tightening shaft in real time and perform closed-loop control with the target values to ensure that the actual tightening process can accurately track the desired torque-angle curve. For example, when the actual tightening angle reaches the end point of a certain segment, the control system will switch to the target torque value of the next segment. If during the tightening process, the actual torque is close to or exceeds the timing torque limit set in step S331, the control system will take protective measures, such as reducing the torque output or stopping the tightening operation, to avoid damaging the connecting parts. Finally, for each bolt connection, a complete set of six-degree-of-freedom robotic arm control instruction sequences is generated. This sequence accurately controls the motion of the robotic arm and the torque output of the end effector, realizing torque-angle compound control based on pre-tightening force misalignment prediction and fatigue risk assessment, and obtaining torque-angle compound control data.

[0061] The present invention also provides a six-degree-of-freedom robotic arm control system for implementing the six-degree-of-freedom robotic arm control method described above. The six-degree-of-freedom robotic arm control system includes: A structural vibration frequency acquisition module, configured to obtain a schematic diagram of automotive component assembly; perform a multi-robot cooperation process analysis on the schematic diagram of automotive component assembly to obtain a multi-robot cooperation process for component assembly; and based on the multi-robot cooperation process for component assembly, collect the structural vibration frequencies of multi-robot assembly through an acoustic emission sensor installed on the six-degree-of-freedom robotic arm to obtain the vibration frequencies of the assembled body structure with sequential filling. A connection structure fracture prediction module, configured to perform a simulation of the disorder of the vibration direction impact of the connection structure interface based on the vibration frequencies of the assembled body structure with sequential filling to obtain disordered data of the impact force in the vibration direction of the connection interface; perform a regression analysis of the misalignment of the connection pre-tightening force based on the disordered data of the impact force in the vibration direction of the connection interface to obtain regression data of the misalignment of the connection pre-tightening force; and predict the fatigue fracture failure of the connection structure based on the regression data of the misalignment of the connection pre-tightening force to obtain fatigue fracture prediction data of the connection structure. A torque-angle composite control module, configured to perform torque-angle composite control of the six-degree-of-freedom robotic arm based on the regression data of the misalignment of the connection pre-tightening force and the fatigue fracture prediction data of the connection structure to obtain torque-angle control iteration data. A control firmware design module, configured to design the control firmware of the six-degree-of-freedom robotic arm based on the torque-angle control iteration data to obtain torque-angle control firmware; and embed the torque-angle control firmware into the six-degree-of-freedom robotic arm to execute the control of the six-degree-of-freedom robotic arm.

[0062] 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. 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 will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A six-degree-of-freedom robotic arm control method, characterized in that, Including the following steps: Step S1: Obtain the assembly schematic diagram of automotive parts; perform a multi-robot collaboration process analysis on the assembly schematic diagram of automotive parts to obtain the multi-robot collaboration process for parts assembly; Based on the multi-robot collaboration process for parts assembly, and collect the vibration frequencies of the multi-robot assembly structure through the acoustic emission sensors installed on the six-degree-of-freedom robotic arm to obtain the vibration frequencies of the assembly body structure filled in time series; Step S2: Perform a simulation on the disorder of the vibration direction impact at the connection interface of the assembly connection structure based on the vibration frequencies of the assembly body structure filled in time series to obtain the disordered data of the impact force in the vibration direction at the connection interface; perform a misalignment regression analysis on the pre-tightening force of the connection based on the disordered data of the impact force in the vibration direction at the connection interface to obtain the misalignment regression data of the pre-tightening force of the connection; Predict the fatigue fracture failure of the connection structure according to the misalignment regression data of the pre-tightening force of the connection to obtain the prediction data of the fatigue fracture of the connection structure; Step S3: Perform torque-angle composite control of the six-degree-of-freedom robotic arm according to the misalignment regression data of the pre-tightening force of the connection and the prediction data of the fatigue fracture of the connection structure to obtain the iterative data of torque-angle control; Step S4: Design the control firmware of the six-degree-of-freedom robotic arm based on the iterative data of torque-angle control to obtain the torque-angle control firmware; Embed the torque-angle control firmware into the six-degree-of-freedom robotic arm to perform the control of the six-degree-of-freedom robotic arm.

2. The six-degree-of-freedom robotic arm control method according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Obtain the assembly schematic diagram of automotive parts; Step S12: Perform a multi-robot collaboration process analysis on the assembly schematic diagram of automotive parts to obtain the multi-robot collaboration process for parts assembly; Step S13: Based on the multi-robot collaboration process for parts assembly, and collect the vibration frequencies of the multi-robot assembly body structure through the acoustic emission sensors installed on the six-degree-of-freedom robotic arm to obtain the vibration frequencies of the multi-robot assembly body structure; Step S14: Insert time series tags into the vibration frequencies of the multi-robot assembly body structure to obtain the vibration frequencies of the assembly structure in time series; Step S15: Fill in the missing values of the vibration frequencies of the assembly structure in time series to obtain the vibration frequencies of the assembly body structure filled in time series.

3. The six-degree-of-freedom robotic arm control method according to claim 1, wherein Step S2 includes the following steps: Step S21: Perform an asymmetric analysis on the vibration frequencies of the assembly body structure filled in time series to obtain the asymmetric dynamic frequencies of the assembly body structure; Step S22: Perform a simulation on the disorder of the vibration direction impact at the connection interface of the assembly connection structure based on the assembly schematic diagram of automotive parts and the asymmetric dynamic frequencies of the assembly body structure to obtain the disordered data of the impact force in the vibration direction at the connection interface; Step S23: Perform a misalignment regression analysis on the pre-tightening force of the connection based on the assembly schematic diagram of automotive parts and the disordered data of the impact force in the vibration direction at the connection interface to obtain the misalignment regression data of the pre-tightening force of the connection; Step S24: Predict the fatigue fracture failure of the connection structure according to the disordered data of the impact force in the vibration direction at the connection interface and the misalignment regression data of the pre-tightening force of the connection to obtain the prediction data of the fatigue fracture of the connection structure.

4. The six-degree-of-freedom robotic arm control method according to claim 3, wherein, Step S22 includes the following steps: Step S221: Extract the morphological structure of the connecting parts from the assembly schematic diagram of automotive parts to obtain the morphological structure data of the connecting parts; Step S222: Perform a coupling analysis of the vibration transmission of the connecting piece on the morphological structure data of the connecting piece according to the asymmetric dynamic frequency of the assembled body structure to obtain the vibration transmission coupling data of the connecting piece; Step S223: Fit the random process of the vibration direction of the connecting piece to the vibration transmission coupling data of the connecting piece to obtain the random process data of the vibration direction of the connecting piece; Step S224: Based on the random process data of the vibration direction of the connecting piece, the vibration transmission coupling data of the connecting piece, and the asymmetric dynamic frequency of the assembled body structure, perform a superposition fitting of the impact force in the direction of the connection interface to obtain the superposition data of the impact force in the direction of the connection interface; Step S225: According to the superposition data of the impact force in the direction of the connection interface, perform a simulation of the disorder of the impact in the vibration direction of the assembled connection structure interface to obtain the disorder data of the impact in the vibration direction of the connection interface.

5. The six-degree-of-freedom robotic arm control method according to claim 4, characterized in that, Step S224 includes the following steps: Calculate the ratio of the acceleration of the frequency response of the vibration transmission coupling data of the connecting piece to obtain the ratio of the acceleration of the vibration frequency of the connecting piece; Based on the random process data of the vibration direction of the connecting piece, perform a coupling process on the ratio of the acceleration of the vibration frequency of the connecting piece to the multi-directional frequency diffusion effect to obtain the coupling data of the multi-directional frequency diffusion effect; Perform an interpolation process on the diffusion radius of the frequency vibration energy intensity of the coupling data of the multi-directional frequency diffusion effect to obtain the interpolation data of the diffusion radius of the frequency intensity; Based on the interpolation data of the diffusion radius of the frequency intensity and the asymmetric dynamic frequency of the assembled body structure, identify the equal gradient of the spatial energy density vector between the connection interfaces to obtain the equal gradient of the spatial energy density vector; Based on the equal gradient of the spatial energy density vector, perform a superposition fitting of the impact force in the direction of the connection interface to obtain the superposition data of the impact force in the direction of the connection interface.

6. The six-degree-of-freedom robotic arm control method according to claim 3, wherein Step S23 includes the following steps: Step S231: Extract the theoretical pre-tightening force of the assembly connection torque based on the schematic diagram of the assembly of automotive parts to obtain the theoretical pre-tightening force of the assembly connection torque; Step S232: Perform a multi-scale analysis of the impact force in the azimuth vibration direction on the disordered data of the impact force in the azimuth vibration direction of the connection interface to obtain the multi-scale data of the disordered impact force in the azimuth vibration; Step S233: Analyze the difference in the skewness of the impact force fluctuation in the spatial dimension on the multi-scale data of the disordered impact force in the azimuth vibration to obtain the difference data of the skewness of the impact force fluctuation in the space; Step S234: According to the difference data of the skewness of the impact force fluctuation in the space, perform a simulation identification of the gradient of the pre-tightening force interference misalignment on the theoretical pre-tightening force of the assembly connection torque to obtain the gradient data of the pre-tightening force interference misalignment; Step S235: Perform a regression analysis of the misalignment of the connection pre-tightening force on the gradient data of the pre-tightening force interference misalignment to obtain the regression data of the misalignment of the connection pre-tightening force.

7. The six-degree-of-freedom robotic arm control method according to claim 6, characterized in that Step S234 includes the following steps: Perform an equal ratio calculation of the difference in the skewness and kurtosis of the spatial azimuth on the difference data of the skewness of the impact force fluctuation in the space to obtain the equal ratio data of the difference in the skewness and kurtosis of the azimuth; Based on the equal ratio data of the difference in the skewness and kurtosis of the azimuth, perform a simulation evaluation of the slip amount of the pre-tightening friction torque on the theoretical pre-tightening force of the assembly connection torque to obtain the data of the slip amount of the pre-tightening friction; Based on the data of the slip amount of the pre-tightening friction and the equal ratio data of the difference in the skewness and kurtosis of the azimuth, perform an analysis of the mismatch distribution of the local pre-tightening force mode of the connection surface to obtain the mismatch distribution data of the local pre-tightening force. Based on the local pre-tightening force mismatch distribution data, perform pre-tightening force interference misalignment gradient simulation identification to obtain pre-tightening force interference misalignment gradient data.

8. The six-degree-of-freedom robotic arm control method according to claim 1, characterized in that Step S3 includes the following steps: Step S31: Perform convolution processing on the connection pre-tightening force misalignment regression data to obtain connection pre-tightening force misalignment convolution data; Step S32: Perform feature structure analysis on the connection structure fatigue fracture prediction data to obtain structure fatigue fracture feature data; Step S33: According to the connection pre-tightening force misalignment convolution data and the structure fatigue fracture feature data, perform torque-angle composite control of the six-degree-of-freedom robotic arm to generate torque-angle composite control data; Step S34: Perform iterative learning on the torque-angle composite control data to obtain torque-angle control iterative data.

9. The six-degree-of-freedom robotic arm control method according to claim 7, wherein Step S33 includes the following steps: Step S331: According to the connection pre-tightening force misalignment convolution data and the structure fatigue fracture feature data, perform connection component timing torque limit matching to obtain connection component timing torque limit matching data; Step S332: Based on the connection pre-tightening force misalignment convolution data, perform connection component screwing angle correction perception to obtain screwing angle correction perception data; Step S333: According to the connection component timing torque limit matching data, perform segmented torque application matching on the screwing angle correction perception data to obtain angle-related torque segmented application data; Step S334: According to the angle-related torque segmented application data, perform torque-angle composite control of the six-degree-of-freedom robotic arm to generate torque-angle composite control data.

10. A six-degree-of-freedom robotic arm control system, characterized in that, For implementing the six-degree-of-freedom robotic arm control method as described in claim 1, the six-degree-of-freedom robotic arm control system includes: A structure vibration frequency acquisition module, configured to obtain an assembly schematic diagram of automotive parts; perform multi-robot cooperation process analysis on the assembly schematic diagram of automotive parts to obtain a multi-robot cooperation process for parts assembly; based on the multi-robot cooperation process for parts assembly, and through an acoustic emission sensor assembled on the six-degree-of-freedom robotic arm, perform multi-robot assembly structure vibration frequency acquisition to obtain the vibration frequency of the assembly body structure in chronological order; A connection structure fracture prediction module, configured to perform simulation of the disorder of the vibration azimuth impact of the assembly connection structure interface based on the vibration frequency of the assembly body structure in chronological order to obtain disordered data of the vibration azimuth impact force of the connection interface; perform connection pre-tightening force misalignment regression analysis based on the disordered data of the vibration azimuth impact force of the connection interface to obtain connection pre-tightening force misalignment regression data; perform connection structure fatigue fracture failure prediction according to the connection pre-tightening force misalignment regression data to obtain connection structure fatigue fracture prediction data; A torque-angle composite control module, configured to perform torque-angle composite control of the six-degree-of-freedom robotic arm according to the connection pre-tightening force misalignment regression data and the connection structure fatigue fracture prediction data to obtain torque-angle control iterative data; A control firmware design module, configured to perform six-degree-of-freedom robotic arm control firmware design based on the torque-angle control iterative data to obtain torque-angle control firmware; embed the torque-angle control firmware into the six-degree-of-freedom robotic arm to execute six-degree-of-freedom robotic arm control.

Images