Car body flexible welding process and related equipment

Abstract

The invention discloses a vehicle body flexible welding process and related equipment. The process comprises five steps of parameter space real-time partitioning, initial control strategy generation, key moment dynamic correction, group collaborative optimization and adaptive parameter adjustment. Identifying a core control parameter and a compensation parameter through scientific partitioning of a multi-dimensional parameter space; retrieving an optimal control strategy based on the physical feature vector; identifying a key control moment by utilizing energy gradient analysis to implement accurate intervention; considering a welding spot group stress transfer relation to optimize an execution sequence; and performing closed-loop correction on the whole control link according to the real-time feedback. According to the technical scheme, the problem of complex coupling of the welding parameters in the multi-vehicle-type high-frequency switching environment can be effectively solved, dynamic optimization and rapid convergence of the nonlinear relation between the parameters are achieved, and the welding quality consistency and the resource utilization efficiency are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent resource scheduling for flexible body welding, and particularly to a flexible body welding process and related equipment. Background Art

[0002] Flexible body welding is an advanced production process in modern automobile manufacturing. It realizes the welding operation of multiple vehicle body models on the same production line through programmable automation equipment and intelligent robot systems. This technology breaks through the limitation of traditional rigid production lines that can only produce a single vehicle model, enabling manufacturers to quickly adjust production plans according to market demands, and improving equipment utilization rate and production efficiency. The flexible welding system mainly consists of multi-joint robots, programmable controllers, intelligent fixture systems, and quick-change tooling, etc., which can adapt to the welding requirements of different vehicle models and achieve flexible production of multiple varieties and small batches.

[0003] While the existing flexible body welding process realizes rapid switching between different vehicle models, it faces severe challenges in the stability of welding quality. Traditional welding processes mainly rely on discrete empirical values and static parameter tables, and cannot fully handle the complex coupling relationships and their dynamic changes among multi-dimensional welding parameters (such as current, pressure, time, material state). When the production line frequently switches between different vehicle models and sheet metal combinations, the non-linear interaction between these parameters will significantly affect the quality of welding spots. Especially in a high-frequency switching environment, the slight fluctuations in the surface state of the material will further complicate the coupling relationship between parameters, and conventional welding methods are difficult to cope with this complex change, resulting in fluctuations in welding spot quality, excessive wear of electrodes, and energy waste. Based on this, there is an urgent need for those skilled in the art to make improvements in this regard. Summary of the Invention

[0004] The main purpose of the present invention is to develop an intelligent process system for flexible body welding, solve the problem of quality stable control under the coupling of multi-dimensional welding parameters, realize the dynamic optimization and rapid convergence of complex non-linear relationships between parameters, and ensure the consistency of welding quality and the maximization of resource utilization efficiency under high-frequency vehicle model switching.

[0005] The first aspect of the present invention provides a flexible body welding process for a vehicle body, and the flexible body welding process for a vehicle body includes: Obtaining system operation parameter data, calculating the change rate and response characteristics of the operation parameters, partitioning the parameter space in real time according to the change rate and the response characteristics, dividing the parameter space into a core control parameter group, a steady-state balance parameter group, and a compensation adjustment parameter group, and generating parameter partition data; Extract the system state feature vector according to the core control parameter group in the parameter partition data, calculate the deviation value between the state feature vector and the preset control target, retrieve a similar control strategy from the control experience library based on the deviation value, and generate an initial control parameter group; Collect the real-time feedback data of the system, calculate the dynamic response characteristics of the system based on the initial control parameter group, identify the key control moments of the system response according to the dynamic response characteristics, and dynamically correct the initial control parameter group at the key control moments to output a control instruction sequence; Input the control instruction sequence into the group control network, calculate the coupling influence matrix between control objects, and co-optimize the control instruction sequence according to the coupling influence matrix to generate a group optimization control strategy; Execute the group optimization control strategy and collect the real-time control feedback data, calculate the real-time control deviation, and perform online correction on the parameter partition data, the initial control parameter group, the control instruction sequence, and the group optimization control strategy according to the real-time control deviation to achieve adaptive adjustment of the control parameters.

[0006] Preferably, the obtaining of the system operation parameter data, calculating the change rate and response characteristics of the operation parameters, partitioning the parameter space in real time according to the change rate and the response characteristics, and dividing the parameter space into a core control parameter group, a steady-state balance parameter group, and a compensation adjustment parameter group to generate parameter partition data includes: Obtain the operation parameter data of the heat input intensity, the electrode pressure, and the position compensation amount, calculate the parameter change rate according to the operation parameter data, perform multi-point sampling on the parameter change rate, and generate a parameter dynamic sequence; Calculate the ratio of the parameter response time to the material softening time based on the parameter dynamic sequence, use the ratio as the parameter time series characteristic, and construct a parameter response function according to the parameter time series characteristic to generate a parameter response characteristic; Perform cross-mapping analysis on the parameter dynamic sequence and the parameter response characteristic, calculate the causal association strength between parameters, construct a parameter link network based on the causal association strength, and generate a parameter coupling relationship; Identify the key nodes of the parameter control chain based on the parameter coupling relationship, calculate the fan-in coefficient and the fan-out coefficient of the key nodes, evaluate the parameter influence according to the fan-in coefficient and the fan-out coefficient, and generate a parameter importance; Classify the parameters according to the parameter importance, classify the parameters with an influence higher than the first threshold into the core control parameter group, classify the parameters with an influence between the first threshold and the second threshold into the steady-state balance parameter group, and classify the parameters with an influence lower than the second threshold into the compensation adjustment parameter group to generate parameter partition data.

[0007] Preferably, extracting a system state feature vector according to the core control parameter group in the parameter partition data, calculating a deviation value between the state feature vector and a preset control target, and retrieving a similar control strategy from a control experience library based on the deviation value to generate an initial control parameter group, including: Extracting welding current density and pressure distribution data according to the core control parameter group in the parameter partition data, calculating a predicted value of the nugget size according to the welding current density and the pressure distribution data, and constructing a nugget growth feature in combination with the solder joint position information to generate a state feature component; Obtaining a resistance change curve between the electrode and the workpiece, performing segmented processing on the resistance change curve, and calculating a slope sequence of the resistance change curve in different welding stages to generate a dynamic response component; Orthogonally combining the state feature component and the dynamic response component to construct a state feature vector reflecting the physical essence of the welding process, and calculating a deviation value of the state feature vector in a multi-dimensional process space according to a preset welding quality target; Establishing a welding condition similarity evaluation function according to the deviation value, retrieving a similar welding condition from a control experience library based on the welding condition similarity evaluation function, and extracting a parameter configuration corresponding to the similar welding condition to generate a candidate control strategy; Performing a process window verification on the candidate control strategy, calculating a process feasibility of the candidate control strategy, and screening a parameter configuration that meets process constraints according to the process feasibility to generate an initial control parameter group.

[0008] Preferably, the acquisition system feeds back data in real time, calculates a system dynamic response feature based on the initial control parameter group, identifies a key control moment of the system response according to the dynamic response feature, and dynamically corrects the initial control parameter group at the key control moment to output a control instruction sequence, including: Collecting real-time feedback data of the current waveform, pressure waveform, and electrode displacement amount during the welding process, performing segmented response analysis on the current parameter and pressure parameter in the initial control parameter group, and calculating the energy input efficiency and heat loss ratio of each segment to generate a multi-segment response feature; Calculating an energy balance state of the welding process according to the multi-segment response feature, extracting energy mutation points and thermal equilibrium points during the welding process, and constructing a state transition diagram of the welding process based on the energy mutation points and the thermal equilibrium points to generate a process state sequence; Performing an energy gradient analysis on the process state sequence, identifying demarcation points of the nugget formation period, transition period, and stable period, and calculating a critical energy threshold for state conversion according to the demarcation points to generate a key control moment; Based on the key control moments and the critical energy threshold, a parameter compensation mapping function is established to map the initial control parameter group to the parameter space of different welding stages. The parameter adjustment amount is calculated according to the critical energy threshold, and segmented control parameters are generated. Perform energy continuity verification on the segmented control parameters, calculate the energy jump amount during parameter switching, and perform parameter smoothing compensation according to the energy jump amount, and output a control instruction sequence.

[0009] Preferably, the energy gradient analysis is performed on the process state sequence to identify the demarcation points of the fusion nucleus formation period, the transition period and the stable period. According to the demarcation points, the critical energy threshold for state conversion is calculated, and key control moments are generated, including: Calculate the energy accumulation curve of the process state sequence, perform differential operation on the energy accumulation curve to obtain the energy change rate curve, construct an energy fluctuation index according to the fluctuation characteristics of the energy change rate curve, and generate an energy characteristic sequence. Perform wavelet transform on the energy characteristic sequence, extract the energy mutation characteristics and the energy gradual change characteristics, identify the energy conversion points according to the combination mode of the energy mutation characteristics and the energy gradual change characteristics, and generate a state conversion sequence. Calculate the energy difference between adjacent conversion points according to the state conversion sequence, perform clustering analysis on the energy difference, identify the energy transition characteristics of the fusion nucleus formation period, the transition period and the stable period based on the clustering analysis results, and generate a demarcation point sequence. Perform energy density analysis on the demarcation point sequence, calculate the energy density gradient at each demarcation point, determine the critical energy threshold for state conversion according to the energy density gradient, and generate key control moments.

[0010] Preferably, the control instruction sequence is input into the group control network, the coupling influence matrix between control objects is calculated, and the control instruction sequence is collaboratively optimized according to the coupling influence matrix to generate a group optimization control strategy, including: Obtain the spatial distribution data of the solder joint group and the plate stress field data, perform stress field analysis on the control instruction sequence, calculate the stress transfer chain and the deformation diffusion field between the solder joints, construct a state propagation network reflecting the group dynamic evolution law according to the stress transfer chain and the deformation diffusion field, and generate a group control topology. Analyze the local stability of the solder joint group based on the group control topology, calculate the singular value distribution of the solder joint stress field, identify the stress field fracture risk points and deformation accumulation points according to the singular value distribution, and generate a coupling influence matrix. Perform structural decomposition on the coupling influence matrix, extract the key control modes of the solder joint group, calculate the structural stiffness contribution spectrum of the solder joint group according to the key control modes, group and sort the solder joint execution sequences based on the structural stiffness contribution spectrum, and generate a group cooperation plan; Group and decouple the control instruction sequence according to the group cooperation plan, calculate the energy distribution ratio of each group of solder joints, perform parameter reconstruction based on the energy distribution ratio, and perform compensation optimization in combination with structural strength constraints to generate a group optimization control strategy.

[0011] Preferably, the performing structural decomposition on the coupling influence matrix, extracting the key control modes of the solder joint group, calculating the structural stiffness contribution spectrum of the solder joint group according to the key control modes, grouping and sorting the solder joint execution sequences based on the structural stiffness contribution spectrum, and generating a group cooperation plan includes: Perform singular value decomposition on the coupling influence matrix, extract the eigenvectors corresponding to the main singular values, construct a modal importance index according to the energy distribution characteristics of the eigenvectors, and generate a modal pedigree; Calculate the modal coupling strength between solder joints based on the modal pedigree, perform hierarchical analysis on the modal coupling strength, identify the key control modes according to the hierarchical analysis results, and generate a modal control sequence; Perform stress transfer analysis on the modal control sequence, calculate the contribution rate of each mode to the structural stiffness, construct a stiffness distribution function based on the contribution rate, and generate a stiffness contribution spectrum; Perform hierarchical clustering on the solder joint group according to the stiffness contribution spectrum, calculate the execution priority of each category of solder joints, design the solder joint execution order based on the execution priority, and generate a group cooperation plan.

[0012] Preferably, the performing the group optimization control strategy and collecting real-time control feedback data, calculating the real-time control deviation, and performing online correction on the parameter partition data, the initial control parameter group, the control instruction sequence, and the group optimization control strategy according to the real-time control deviation to achieve adaptive adjustment of the control parameters includes: Perform the group optimization control strategy and collect real-time data of welding current, electrode pressure, and displacement, extract the quality characteristics of the real-time data, calculate the nugget size deviation and the solder joint strength distribution, and generate a control deviation vector; Perform causal decomposition on the control deviation vector, identify the deviation components caused by process parameters, material states, and environmental factors, calculate the influence weights according to the transmission paths of the deviation components, and generate a deviation compensation sequence; Calculate the drift amount of the parameter partition boundary based on the deviation compensation sequence, adjust the boundaries of the parameter partition data and the initial control parameter group, and correct the parameter mapping relationship according to the boundary adjustment amount to generate a corrected parameter group; Map the correction parameter group into the control instruction sequence and the population optimization control strategy, perform gradient update on the control parameters according to process constraints, and perform compensation and correction according to welding quality constraints to achieve adaptive adjustment of the control parameters.

[0013] The second aspect of the present invention provides a flexible body welding device, which includes: A parameter partitioning module, configured to obtain multi-dimensional parameter data and perform partitioning processing to obtain core quality area parameters, stable balance area parameters, and environment adaptation area parameters, respectively calculate the influence weights of each parameter on the process, determine the control priority based on the influence weights, and generate a resource allocation scheme; A working condition mapping module, configured to convert target working condition data into a working condition feature vector, calculate the similarity with historical working condition data according to the working condition feature vector, and output initial control parameters based on the similarity and the resource allocation scheme; A timing control module, configured to obtain process state data based on the initial control parameters, calculate the current change rate and temperature gradient according to the state data, determine the process stage, perform microsecond-level control parameter adjustment on the process stage, generate a dynamic intervention instruction, execute parameter regulation according to the dynamic intervention instruction, and output optimized control parameters; A population cooperation module, configured to obtain multi-point distribution data, calculate the influence relationship according to the multi-point distribution data, construct a network topology structure, analyze the temperature field and stress field distributions based on the network topology structure, generate a regional balance compensation scheme, and perform compensation adjustment on the optimized control parameters according to the regional balance compensation scheme to output global coordination parameters; A material adaptation module, configured to establish a control mode space according to material characteristic data, generate a parameter gradual change sequence based on the control mode space, continuously adjust the global coordination parameters according to the parameter gradual change sequence, and output final control parameters.

[0014] The third aspect of the present invention provides a flexible body welding device, including: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected by a line; the at least one processor calls the instructions in the memory so that the flexible body welding device executes the steps of the above-mentioned flexible body welding process.

[0015] The fourth aspect of the present invention provides a computer-readable storage medium, in which instructions are stored, and when it runs on a computer, it causes the computer to execute the steps of the above-mentioned flexible body welding process.

[0016] The technical solution provided by the embodiments of this application calculates the parameter change rate and response characteristics by obtaining the system operation parameter data, scientifically partitions the parameter space, and divides the complex parameter space into a core control parameter group, a steady-state balance parameter group, and a compensation adjustment parameter group. This hierarchical control strategy effectively reduces the control dimension and solves the curse of dimensionality problem caused by multi-dimensional parameter coupling. By identifying the core parameters that truly play a decisive role, the system can concentrate limited computing resources and control precision on the most critical parameters, improving the control efficiency.

[0017] In the initial control strategy generation link, the system extracts the state feature vector according to the partitioned core control parameter group, calculates the deviation value between it and the preset control target, and retrieves similar control strategies from the control experience library. This case-based reasoning method based on the experience library shortens the system convergence time and lays a foundation for subsequent fine-tuning. By constructing a feature vector that reflects the physical essence of welding, the system can grasp the key characteristics of the welding process from the essence, rather than just staying at the simple matching of surface parameters.

[0018] The dynamic response correction link is the key to solving the non-linear control of welding. The system collects real-time feedback data, calculates the dynamic response characteristics, identifies the key control moments of the system response, and performs precise intervention at these moment points. This dynamic control method based on the analysis of the energy balance state breaks through the limitations of the traditional static parameter table, can cope with dynamic change factors such as the material state and heat conduction characteristics during the welding process, ensures that appropriate control instructions are applied at critical moments, and avoids welding quality fluctuations.

[0019] The group collaborative optimization link solves the complex problem of mutual influence between solder joints. The system inputs the control instruction sequence into the group control network, calculates the coupling influence matrix between the controlled objects, and performs collaborative optimization. This control strategy considering the group behavior of solder joints overcomes the defect of traditional single-point control that does not consider stress transfer and deformation accumulation. By optimizing the execution order of solder joints and the energy distribution ratio, the strength and stability of the overall structure are ensured.

[0020] The adaptive parameter adjustment link realizes the continuous optimization of the system. Execute the group optimization control strategy and collect feedback data, calculate the control deviation, and perform online correction on the entire control link. This closed-loop feedback mechanism enables the system to continuously learn and adapt to the changes in the production environment, perform real-time compensation for interference factors such as material state fluctuations and environmental factor changes, and ensure the long-term stability of welding quality.

[0021] Through the organic combination of these five links, the present invention realizes the precise control of the flexible welding process of the vehicle body. Parameter zoning reduces the system complexity, initial strategy generation accelerates system convergence, dynamic response correction copes with non-linear changes, group collaborative optimization considers the overall performance, and adaptive adjustment ensures long-term stability. This multi-level and whole-process control strategy enables the system to quickly converge to the optimal process parameters in a high-frequency vehicle model switching environment, ensuring the consistency of welding quality. At the same time, through precise energy control and electrode life management, the resource utilization efficiency is maximized, providing a solid technical foundation for the flexible manufacturing of the vehicle body. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on the structures shown in these drawings.

[0023] Figure 1 It is a schematic diagram of an embodiment of the flexible welding process of the vehicle body in an embodiment of the present invention; Figure 2 It is a schematic diagram of an embodiment of the flexible welding device of the vehicle body in an embodiment of the present invention; Figure 3 It is a schematic diagram of an embodiment of the flexible welding equipment of the vehicle body in an embodiment of the present invention.

[0024] The realization, functional features and advantages of the object of the present invention will be further described in conjunction with the embodiments and with reference to the drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0026] It should be noted that if there are directional indications (such as up, down, left, right, front, back...) involved in the embodiments of the present invention, the directional indications are only used to explain the relative position relationship and movement conditions between components in a specific posture (as shown in the drawings). If the specific posture changes, the directional indications will also change accordingly.

[0027] In addition, the descriptions involving "first", "second", etc. in the present invention are only for descriptive purposes and should not be construed as indicating or implying their relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one such feature. In addition, "and / or" throughout the text includes three scenarios. Taking A and / or B as an example, it includes the technical solution of A, the technical solution of B, and the technical solution where both A and B are satisfied simultaneously. In addition, the technical solutions between various embodiments can be combined with each other, which must be based on the ability of those of ordinary skill in the art to implement. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0028] An embodiment of the present application provides a flexible body welding process. Figure 1 It is a flowchart of the flexible body welding process provided by an embodiment of the present application. In this embodiment, the method includes: Please refer to Figure 1 , obtain system operation parameter data, calculate the change rate and response characteristics of the operation parameters, perform real-time partitioning of the parameter space according to the change rate and the response characteristics, divide the parameter space into a core control parameter group, a steady-state balance parameter group, and a compensation adjustment parameter group, and generate parameter partitioning data; In an embodiment of the present invention, the obtaining system operation parameter data, calculating the change rate and response characteristics of the operation parameters, performing real-time partitioning of the parameter space according to the change rate and the response characteristics, dividing the parameter space into a core control parameter group, a steady-state balance parameter group, and a compensation adjustment parameter group, and generating parameter partitioning data includes: Obtain the operation parameter data of heat input intensity, electrode pressure, and position compensation amount, calculate the parameter change rate according to the operation parameter data, perform multi-point sampling on the parameter change rate, and generate a parameter dynamic sequence; Based on the parameter dynamic sequence, calculate the ratio of the parameter response time to the material softening time, use the ratio as the parameter time sequence characteristic, construct a parameter response function according to the parameter time sequence characteristic, and generate a parameter response characteristic; Perform cross-mapping analysis on the parameter dynamic sequence and the parameter response characteristic, calculate the causal association strength between parameters, construct a parameter link network based on the causal association strength, and generate a parameter coupling relationship; Based on the parameter coupling relationship, identify the key nodes of the parameter control chain, calculate the fan-in coefficient and fan-out coefficient of the key nodes, evaluate the parameter influence according to the fan-in coefficient and the fan-out coefficient, and generate a parameter importance; Classify the parameters according to the parameter importance degree, classify the parameters with influence higher than the first threshold into the core control parameter group, classify the parameters with influence between the first threshold and the second threshold into the steady-state balance parameter group, classify the parameters with influence lower than the second threshold into the compensation adjustment parameter group, and generate parameter partition data.

[0029] The following specifically describes the steps involved in the above embodiments: Obtaining the operation parameter data of the heat input intensity, electrode pressure, and position compensation amount is achieved through a sensor network installed on the body flexible welding equipment. The welding controller can be connected to a current sensor to collect the welding current signal, calculate the heat input intensity by measuring the temperature distribution in the welding area through a thermal imager, monitor the change of the electrode pressure by a pressure sensor, and measure the electrode displacement by a laser displacement sensor to obtain the position compensation amount. These sensors collect data in real time at a high sampling frequency (usually 1 kHz), and the signal conditioning and digital processing are performed by a data acquisition unit. For the collected parameter data, calculate the ratio of the change amount within a unit time to the time interval to obtain the parameter change rate. For example, the change rate of the heat input intensity can be expressed as the heat change value within a unit time, and the change rate of the electrode pressure represents the pressure change value within a unit time. After calculating the parameter change rate, multi-point sampling is performed at a uniform time interval (typical value is 0.5 ms) to form a parameter dynamic sequence including timestamps and corresponding parameter change rates. This high-frequency sampling method can capture the minute changes in the welding transient process and provide a detailed data basis for subsequent analysis.

[0030] Analyze the relationship between the parameter response characteristics and material characteristics during the welding process based on the parameter dynamic sequence. First, determine the parameter response time, that is, the time required from the parameter change to the system generating a significant response, which is identified by analyzing the mutation points and stable points in the parameter dynamic sequence. For example, when the welding current changes, the time delay from the current change to the formation of the weld nugget is the response time of the heat input intensity. At the same time, calculate the material softening time according to the material characteristics of the steel plate, that is, the time required for the material to reach the plastic deformation state at a specific temperature. Through thermodynamic analysis, the softening time of ordinary body steel plates is about 8 - 12 ms, and that of high-strength steel is about 10 - 15 ms. Calculate the ratio of the parameter response time to the material softening time, and this ratio reflects the timing matching degree between the parameter change and the material state change. A ratio close to 1 indicates that the parameter change is synchronized with the material state change, less than 1 indicates that the parameter change lags behind the material change, and greater than 1 indicates that the parameter change is ahead of the material change. Based on these ratios, construct a parameter response function to describe the influence law of the parameter change on the material state and generate parameter response characteristics. This analysis method based on the timing relationship can accurately reflect the dynamic interaction relationship between the parameters and the material during the welding process.

[0031] Cross - map analysis is performed on the parameter dynamic sequence and parameter response characteristics, and the convergent cross - mapping (CCM) technique is used to evaluate the causal association between parameters. In specific implementation, a state - space reconstruction is constructed, the ability of parameter X to predict parameter Y and the ability of parameter Y to predict parameter X are calculated, and the causal direction and strength are determined through a prediction skill score. For example, if the prediction ability of the heat input intensity on the nugget size is stronger than the reverse prediction ability, it indicates that the heat input intensity has a significant causal effect on the nugget size. After calculating the causal association strength for all parameter pairs, a parameter link network is constructed, where nodes represent parameters and edges represent the causal association strength. For a body flexible welding system, a typical parameter link network may contain 10 - 15 key parameter nodes and 20 - 30 causal links. The weight of the edge in the network represents the causal strength, and the complex coupling relationship between parameters can be visually displayed through visualization techniques. This network analysis method based on causal inference solves the limitation of traditional correlation analysis that cannot distinguish the causal direction and reveals the true influence mechanism between welding parameters.

[0032] Identify the key nodes of the parameter control chain based on the parameter coupling relationship, and use network analysis algorithms to calculate the network centrality index of the nodes. Calculate the in - degree coefficient (representing the degree of being affected by other parameters) and out - degree coefficient (representing the degree of affecting other parameters) for each parameter node. The in - degree coefficient is calculated as the sum of the edge weights pointing to the node, and the out - degree coefficient is calculated as the sum of the edge weights starting from the node. For example, the out - degree coefficient of the welding current node is usually high, indicating that it has a significant impact on multiple downstream parameters (such as nugget size, melting state). Evaluate the comprehensive influence of each parameter by combining the in - degree and out - degree coefficients, and the influence can be expressed as the weighted average of the ratio of out - degree to in - degree. Parameters with high out - degree and low in - degree are usually the driving factors of the system and have a decisive impact on welding quality. This network topology analysis method can quantitatively evaluate the importance and position of parameters in the entire welding system, providing a scientific basis for parameter classification.

[0033] Scientifically classify the parameters according to their importance, and achieve effective management of the complex parameter space. Determine the influence threshold based on experimental verification and expert knowledge. For typical body welding applications, the first threshold is set to 0.75, and the second threshold is set to 0.35. Parameters with an influence higher than 0.75 are classified into the core control parameter group. These parameters directly determine the welding quality and usually include 3 - 5 parameters such as welding current, electrode pressure, and welding time. Parameters with an influence between 0.75 and 0.35 are classified into the steady-state balance parameter group. These parameters affect the system stability and include 4 - 6 parameters such as electrode cooling water flow rate and clamping force. Parameters with an influence lower than 0.35 are classified into the compensation adjustment parameter group. These parameters are used for fine-tuning and adapting to environmental changes and include 5 - 8 parameters such as electrode position compensation and preheating parameters. For example, in high-strength steel welding applications, although the pulse rise time does not directly determine the formation of the fusion nucleus, it has an important impact on preventing spatter and is suitable for classification into the steady-state balance parameter group. This parameter classification method significantly reduces the complexity of the control system, focuses the control resources on the truly important parameters, and improves the calculation efficiency and control accuracy.

[0034] Please continue to refer to Figure 1 , extract the system state feature vector according to the core control parameter group in the parameter partition data, calculate the deviation value between the state feature vector and the preset control target, retrieve the similar control strategy from the control experience library based on the deviation value, and generate the initial control parameter group; In an embodiment of the present invention, the extracting the system state feature vector according to the core control parameter group in the parameter partition data, calculating the deviation value between the state feature vector and the preset control target, retrieving the similar control strategy from the control experience library based on the deviation value, and generating the initial control parameter group includes: Extract the welding current density and pressure distribution data according to the core control parameter group in the parameter partition data, calculate the predicted value of the fusion nucleus size according to the welding current density and the pressure distribution data, combine the solder joint position information to construct the fusion nucleus growth feature, and generate the state feature component; Obtain the resistance change curve between the electrode and the workpiece, perform segmented processing on the resistance change curve, calculate the slope sequence of the resistance change curve in different welding stages, and generate the dynamic response component; Orthogonally combine the state feature component and the dynamic response component to construct a state feature vector reflecting the physical essence of the welding process, and calculate the deviation value of the state feature vector in the multi-dimensional process space according to the preset welding quality target; Establish a welding condition similarity evaluation function according to the deviation value, retrieve the similar welding conditions from the control experience library based on the welding condition similarity evaluation function, extract the parameter configuration corresponding to the similar welding conditions, and generate the candidate control strategy; Perform process window verification on the candidate control strategy, calculate the process feasibility of the candidate control strategy, screen parameter configurations that meet process constraints according to the process feasibility, and generate an initial control parameter set.

[0035] The following specifically describes the steps involved in the above embodiments: When extracting welding current density and pressure distribution data according to the core control parameter set in the parameter partition data, a high-precision data acquisition system is used for real-time monitoring. The welding current density is calculated by measuring the welding current value through a Hall current sensor and combining the electrode contact area, with the unit of A / mm²; the pressure distribution data is obtained through a thin-film pressure sensor array to obtain the pressure distribution on the electrode contact surface. These two groups of key data are input into a nugget prediction model based on finite element analysis, which takes into account the thermophysical properties of the material and the welding geometric conditions, and calculates the predicted nugget size value. For example, when welding a high-strength steel plate with a thickness of 1.2 mm, when the current density is 320 A / mm² and the pressure distribution uniformity coefficient is greater than 0.85, the predicted nugget diameter is about 5.8 mm. Combining with the three-dimensional coordinate data of the solder joints, the spatio-temporal distribution characteristics describing the formation and growth process of the nugget are established, that is, the nugget growth characteristics. This characteristic includes key indicators such as the nugget diameter growth rate, the nugget shape coefficient, and the penetration ratio, forming a state characteristic component. This step builds a bridge connecting process parameters and the welding physical process, ensuring that the control system can accurately grasp the physical essence of the welding process, rather than relying solely on surface parameters.

[0036] Obtaining the resistance change curve between the electrode and the workpiece is achieved through a high-speed resistance measurement system, which records the dynamic change of the resistance between the electrodes during welding at a sampling frequency above 10 kHz. The resistance change curve is an important signal reflecting the change of the physical state of the welding process, and it needs to be segmented into five key stages: the contact stage, the heating stage, the nugget formation stage, the nugget growth stage, and the cooling and solidification stage. Using change point detection algorithms, such as abrupt change point detection or wavelet analysis methods, to identify the demarcation points of each stage. Calculate the slope value of the resistance change curve for each stage to form a slope sequence. For example, for the welding of the B-pillar reinforcement plate and the side panel, the slope of the resistance curve in the nugget formation stage is usually negative (-0.2 to -0.5 mΩ / ms), indicating the start of nugget formation; while in the stable nugget growth stage, the slope tends to be flat (-0.05 to 0.1 mΩ / ms). This set of slope sequences constitutes a dynamic response component, which can sensitively reflect the physical state changes during the welding process. This segmented analysis method overcomes the limitations of traditional single resistance value monitoring and can capture the key state transition points during the welding process.

[0037] The orthogonal combination of the state feature component and the dynamic response component is achieved through a feature fusion algorithm. First, the two components are normalized to eliminate the dimensional difference. Then, the orthogonal feature vectors are extracted by the principal component analysis (PCA) or singular value decomposition (SVD) method to remove redundant information. Finally, combined with the knowledge of welding physics, the features are weighted and combined to construct a state feature vector that reflects the physical essence of the welding process. For example, for the dissimilar material welding of aluminum alloy and steel plate, the state feature vector includes combined features such as "nucleus diameter - resistance drop rate" and "pressure distribution uniformity - contact resistance stability", which can comprehensively describe the formation state of the interfacial intermetallic compound. According to the preset welding quality targets (such as tensile strength greater than 280 MPa and shear strength greater than 200 MPa), the Euclidean distance or Mahalanobis distance between the current state feature vector and the ideal state is calculated to obtain the deviation value in the multi-dimensional process space. This state description method driven by physical features enables the control system to essentially grasp the dynamic evolution law of the welding process.

[0038] The establishment of the welding working condition similarity evaluation function based on the deviation value is achieved through a weighted distance measurement method. This function comprehensively considers three dimensions: material combination similarity, geometric structure similarity, and process parameter similarity, and different weight coefficients are set for each dimension. For example, for the welding points at the same structural position of different vehicle models, the weight of the material combination similarity is set to 0.5, the weight of the geometric structure similarity is 0.3, and the weight of the process parameter similarity is 0.2. Based on this evaluation function, similar working conditions are retrieved from the control experience database containing tens of thousands of groups of historical welding data. The control experience database is a knowledge base continuously accumulated through production practice and offline optimization, and each working condition record includes material information, structural information, process parameters, and quality evaluation results. The similarity threshold is set above 0.85 to ensure the reliability of the retrieval results. The parameter configurations corresponding to the similar working conditions are extracted, including current waveform parameters, pressure control parameters, and timing control parameters, etc., to form candidate control strategies. This case-based reasoning method based on the experience database significantly reduces the parameter optimization time for new working conditions and improves the speed of the system to adapt to new vehicle models.

[0039] The process window verification of the candidate control strategy is achieved through a welding process simulation system and constraint condition checking. First, process constraints are set, including spatter limits (current rise rate less than 200 A / ms), electrode indentation limits (indentation depth less than 0.1 mm), and nugget size requirements (diameter greater than 4.5 mm), etc.; then, the candidate control strategy is numerically simulated using a thermo-electro-mechanical coupling model to predict the welding results under various disturbance conditions; finally, the process feasibility of the strategy is calculated, that is, the probability of meeting the constraint conditions. The process feasibility is calculated using the Monte Carlo method to simulate the welding stability when parameters such as material thickness and surface condition fluctuate within a certain range. The process feasibility threshold is set at 90%, that is, the welding quality can be guaranteed under more than 90% of the disturbance conditions. Eligible parameter configurations are selected according to the calculation results to form an initial control parameter group. This verification method based on the process window ensures the stability and robustness of the initial control strategy in the actual production environment and effectively prevents the occurrence of poor welding states.

[0040] Please continue to refer to Figure 1 , the acquisition system collects real-time feedback data, calculates the system dynamic response characteristics based on the initial control parameter group, identifies the key control moments of the system response according to the dynamic response characteristics, and dynamically corrects the initial control parameter group at the key control moments to output a control instruction sequence; In an embodiment of the present invention, the acquisition system collects real-time feedback data, calculates the system dynamic response characteristics based on the initial control parameter group, identifies the key control moments of the system response according to the dynamic response characteristics, and dynamically corrects the initial control parameter group at the key control moments to output a control instruction sequence, including: Collect real-time feedback data of the current waveform, pressure waveform, and electrode displacement amount during the welding process, perform segmented response analysis on the current parameters and pressure parameters in the initial control parameter group, calculate the energy input efficiency and heat loss ratio of each segment, and generate multi-segment response characteristics; Calculate the energy balance state of the welding process according to the multi-segment response characteristics, extract the energy mutation points and thermal equilibrium points during the welding process, construct a state transition diagram of the welding process based on the energy mutation points and the thermal equilibrium points, and generate a process state sequence; Perform energy gradient analysis on the process state sequence, identify the demarcation points of the nugget formation period, transition period, and stable period, calculate the critical energy threshold for state transition according to the demarcation points, and generate key control moments; Establish a parameter compensation mapping function based on the key control moments and the critical energy threshold, map the initial control parameter group to the parameter space of different welding stages, calculate the parameter adjustment amount according to the critical energy threshold, and generate segmented control parameters; Verify the energy continuity of the segmented control parameters, calculate the energy jump amount during parameter switching, perform parameter smoothing compensation according to the energy jump amount, and output a control instruction sequence.

[0041] The following specifically describes the steps involved in the above embodiments: The real-time feedback data of the current waveform, pressure waveform, and electrode displacement amount during the welding process is obtained through a high-precision sensor system. The welding controller is connected to a high-speed current sensor (sampling rate 20 kHz) to collect the current waveform, a pressure sensor (accuracy ±0.5%) to monitor the change in electrode pressure, and a laser displacement sensor (resolution 0.1 μm) to measure the electrode displacement amount. Perform segmented response analysis on the current parameters and pressure parameters in the initial control parameter group, and divide the welding process into four key segments: the extrusion stage, the preheating stage, the main heating stage, and the holding stage. Calculate the energy input efficiency (i.e., the ratio of input electrical energy converted into effective heat energy) and the heat loss ratio (i.e., the ratio of dissipated heat to total heat) for each stage. The calculation method is to obtain the input power through the product of current and voltage, and then calculate the heat loss in combination with the temperature difference of the electrode cooling water, and finally obtain the energy utilization status of each stage. For example, in the welding condition of connecting the B-pillar of the vehicle body to the side panel, the energy input efficiency in the preheating stage is about 65%, and the heat loss ratio is about 35%; while the energy input efficiency in the main heating stage increases to 78%, and the heat loss ratio drops to 22%. This segmented response analysis method overcomes the limitation of the traditional welding control that treats the entire process as a black box, and realizes the fine grasp of the welding energy flow process.

[0042] Calculate the energy balance state of the welding process according to the multi-segment response characteristics, and analyze the relationship between energy input, accumulation, and dissipation in the welding area using the principle of thermodynamic equilibrium. The calculation method is to establish an energy integration model to track the change rate of energy over time in the welding area. When the difference between the energy input rate and the dissipation rate shows a significant change (change rate greater than 20%), it is identified as an energy mutation point; when the energy input rate and the dissipation rate reach equilibrium (difference less than 5%), it is identified as a thermal equilibrium point. For example, during the welding of high-strength steel plates, the first energy mutation point appears at about 8 ms after the start of welding, indicating that the surface contact resistance of the workpiece is stable; the second energy mutation point appears at about 25 ms, indicating that the weld nugget begins to form; and the thermal equilibrium point is reached at about 60 ms, indicating that the weld nugget grows stably. Based on these characteristic points, construct a state transition diagram of the welding process, use nodes to represent the welding states, and use edges to represent the state transition conditions, generating a process state sequence containing time information. This method based on energy balance analysis can accurately capture the key physical state changes during the welding process, avoiding the control blind spots caused by the traditional method relying on an empirical parameter table.

[0043] Perform energy gradient analysis on the process status sequence, calculate the energy change rate between adjacent status points, and apply the gradient clustering algorithm to group status points with similar energy change patterns into the same stage. By identifying significant change points of the energy gradient (gradient change greater than 30%), determine the demarcation points of the nugget formation period (rapid energy accumulation period, gradient value greater than 15 J / ms), the transition period (period of slow energy growth, gradient value between 5 - 15 J / ms), and the stable period (energy balance period, gradient value less than 5 J / ms). For example, during the welding process of the outer panel and inner panel of a car door, the demarcation point between the nugget formation period and the transition period appears at 28.5 ms of the welding time, at which time the nugget diameter reaches 60% of the plate thickness; the demarcation point between the transition period and the stable period appears at 72.3 ms, at which time the nugget has formed a stable structure. Calculate the critical energy threshold for state transition based on these demarcation points, that is, the minimum energy value required to trigger state transition. For cold-rolled steel plates with a thickness of 1.2 mm, the critical energy threshold from the solid state to the nugget formation period is approximately 450 J, and the critical energy threshold from the nugget formation period to the stable period is approximately 1200 J. These critical time points and energy thresholds provide a time window for precise control, enabling the system to intervene at the most effective moment.

[0044] Establish a parameter compensation mapping function based on the critical control moments and critical energy thresholds. This function maps the general parameters in the initial control parameter group to the specific parameter spaces of different welding stages. The mapping function achieves smooth transition through the hyperbolic tangent function to ensure the continuity of parameter changes. Design differentiated control strategies according to the characteristics of different welding stages: focus on quickly providing sufficient energy during the nugget formation period, with the current parameter increased by 8 - 15%; focus on stabilizing the nugget growth during the transition period, with the pressure parameter finely adjusted by ±5%; focus on maintaining the nugget quality during the stable period, with the current parameter decreased by 5 - 10%. For example, for the welding condition of the body longitudinal beam, the initial current parameter is 8.5 kA, which is mapped to 9.2 kA during the nugget formation period, adjusted to 8.7 kA during the transition period, and reduced to 8.1 kA during the stable period. Calculate the parameter adjustment amount according to the critical energy threshold to ensure that the energy input in each stage matches the physical requirements. The generated segmented control parameters include the current value, pressure value, and their action time in each stage, forming a complete segmented control scheme. This parameter regulation method based on physical stages breaks through the limitation of using a single parameter group in traditional welding and realizes precise control of the entire welding process.

[0045] Verify the energy continuity of the segmented control parameters to ensure that there is no sudden change in energy during parameter switching. The calculation method is to calculate the difference in energy input per unit time before and after switching at each parameter switching point, that is, the energy jump amount. For high-quality welding, controlling the energy jump amount within the range of ±10% can ensure stable welding quality. When it is detected that the energy jump amount exceeds the threshold, apply a parameter smoothing compensation algorithm to reduce the severity of energy change by adjusting the slope of parameter switching or inserting transitional section parameters. For example, when transitioning from the fusion nucleus formation period to the stable period, if directly reducing the current from 9.5 kA to 8.2 kA will cause a sudden change in energy input, the system will automatically insert a 50-ms gradual change process to smoothly reduce the current. The finally output control instruction sequence contains complete timing information and parameter values, and is transmitted as a digital signal to the welding controller for execution. This parameter optimization method based on energy continuity avoids quality problems such as spatter and cracks caused by sudden parameter changes in traditional welding control, and improves the stability of the welding process and the consistency of solder joint quality.

[0046] In an embodiment of the present invention, the energy gradient analysis of the process state sequence is performed to identify the demarcation points of the fusion nucleus formation period, the transition period, and the stable period, and the critical energy threshold for state transition is calculated according to the demarcation points to generate key control moments, including: Calculate the energy accumulation curve of the process state sequence, perform a differential operation on the energy accumulation curve to obtain the energy change rate curve, construct an energy fluctuation index according to the fluctuation characteristics of the energy change rate curve, and generate an energy characteristic sequence; Perform wavelet transform on the energy characteristic sequence, extract the energy sudden change characteristics and energy gradual change characteristics, identify the energy conversion points according to the combined mode of the energy sudden change characteristics and the energy gradual change characteristics, and generate a state transition sequence; Calculate the energy difference between adjacent conversion points according to the state transition sequence, perform cluster analysis on the energy difference, identify the energy transition characteristics of the fusion nucleus formation period, the transition period, and the stable period based on the results of the cluster analysis, and generate a demarcation point sequence; Perform energy density analysis on the demarcation point sequence, calculate the energy density gradient at each demarcation point, determine the critical energy threshold for state transition according to the energy density gradient, and generate key control moments.

[0047] The following specifically describes the steps involved in the above embodiment: The energy accumulation curve of the calculation process status sequence is achieved by integrating the current, voltage, and time data during the welding process. The specific operation is to use a high-speed data acquisition system to collect welding power data, and the sampling frequency is set to 20 kHz to ensure capturing energy changes at the microsecond level. The cumulative energy calculation formula is E(t)=∫P(τ)dτ, where P(τ) is the instantaneous power at time τ, and the integration interval is from the start of welding to time t. Perform five-point sliding differentiation on the obtained energy accumulation curve E(t) to obtain the energy change rate curve dE(t) / dt, which intuitively reflects the dynamic change of the energy input rate during the welding process. Construct an energy fluctuation index based on the fluctuation characteristics of the energy change rate curve. The calculation method is the ratio of the standard deviation to the mean of the energy change rate within a continuous 5 ms window. For example, during the welding of high-strength steel for the A-pillar of the vehicle body, the energy fluctuation index in the interval of 15 - 20 ms after the start of welding is 0.08, indicating stable energy input; while the energy fluctuation index in the interval of 22 - 27 ms rises to 0.35, indicating entry into the material phase change stage. Combine the time series with the corresponding energy fluctuation index to generate an energy characteristic sequence. This processing method converts the complex welding energy process into numerical characteristics that are convenient for analysis, enabling the system to accurately capture the minute changes in the energy input state.

[0048] Performing wavelet transform on the energy characteristic sequence uses the Daubechies wavelet (db4) for multi-scale decomposition, and the decomposition level is set to 5 layers, which can capture the energy change characteristics at different time scales simultaneously. The advantage of wavelet transform is that it can extract the time-domain and frequency-domain information of the signal simultaneously, and is particularly suitable for analyzing transient changes during the welding process. By analyzing the amplitude and distribution characteristics of the wavelet coefficients, two types of key characteristics are extracted: energy mutation characteristics (the coefficient amplitude suddenly increases and the duration is short, and the amplitude change is greater than 200%) and energy gradual change characteristics (the coefficient amplitude changes slowly and the duration is long, and the amplitude change is between 50% - 100%). Identify the energy conversion points according to the combination mode of these two types of characteristics: the mutation characteristic followed by the gradual change characteristic usually indicates entry into a new welding stage. For example, during the welding of the vehicle body sill panel, a significant mutation characteristic appears at 26.5 ms after the start of welding, and then enters the gradual change characteristic, marking the transition from solid-state heating to the initial formation stage of the weld nugget. Arrange all the identified energy conversion points in chronological order to generate a state transition sequence. This feature extraction method based on wavelet analysis can accurately identify the key change points of the energy state, without being disturbed by noise and minor fluctuations.

[0049] Calculate the energy difference between adjacent transition points according to the state transition sequence, that is, the total energy absorbed by the welding system between every two adjacent transition points. The calculation method is to integrate and find the difference of the energy accumulation curve between two transition points. Apply the K-means clustering algorithm (set K value to 3) to the obtained energy difference sequence, and cluster the energy differences into three categories: low energy interval (corresponding to the surface contact stage), medium energy interval (corresponding to the nugget formation stage), and high energy interval (corresponding to the stable growth stage of the nugget). Based on the clustering results, identify the energy transition characteristics of the three key stages: the characteristic of the nugget formation period is that the energy difference jumps from the low energy interval to the medium energy interval, usually with an increase in cumulative energy of 300 - 500 J; the characteristic of the transition period is that the energy difference fluctuates within the medium energy interval, with an increase in cumulative energy of 500 - 800 J; the characteristic of the stable period is that the energy difference enters the high energy interval and the fluctuation decreases, with an increase in cumulative energy of 800 - 1200 J. For example, for the welding of medium and high strength steel, the energy transition value from surface contact to nugget formation is 420 J, and the energy transition value from nugget formation to the stable period is 650 J. Determine the demarcation points of the key stages according to these energy transition characteristics, and generate a demarcation point sequence containing accurate timestamps. This method of stage identification based on energy transition is directly based on the essential characteristics of the welding physical process, overcoming the limitations of traditional division methods based on time or surface phenomena.

[0050] Conduct energy density analysis on the demarcation point sequence. First, calculate the energy density (energy content per unit volume) at each demarcation point. The calculation method is to divide the cumulative energy at the demarcation point by the estimated affected volume (related to the electrode contact area and the depth of the heat affected zone). Then calculate the energy density gradient, that is, the change rate of the energy density per unit time. For typical body welding applications, the energy density gradient at the first demarcation point (from contact to nugget formation) is about 4 - 6 J / (mm³·ms), and the energy density gradient at the second demarcation point (from nugget formation to stable growth) is about 1.5 - 3 J / (mm³·ms). Determine the critical energy thresholds for state transition according to these energy density gradients: the critical energy density threshold for nugget formation is set to 28 - 32 J / mm³, and the critical energy density threshold for transitioning to the stable period is set to 45 - 55 J / mm³. These critical thresholds are closely related to the melting point, specific heat capacity, and latent heat of the material. For example, for low carbon steel with a carbon content of 0.15%, the theoretical energy density required for melting is about 30 J / mm³, which is consistent with the experimentally determined threshold. Determine the time points corresponding to these critical energy thresholds as the key control moments, and form an accurate intervention schedule for the welding process. This method based on energy density analysis establishes a direct connection between the welding physical state and the control moment, enabling the control system to implement precise control at the most appropriate time and meet the requirements of different material combinations and welding conditions.

[0051] Please continue to refer to Figure 1, input the control instruction sequence into the group control network, calculate the coupling influence matrix between the control objects, and perform collaborative optimization on the control instruction sequence according to the coupling influence matrix to generate a group optimization control strategy; In an embodiment of the present invention, the step of inputting the control instruction sequence into the group control network, calculating the coupling influence matrix between the control objects, performing collaborative optimization on the control instruction sequence according to the coupling influence matrix, and generating a group optimization control strategy includes: Obtain the spatial distribution data of the solder joint group and the plate stress field data, perform stress field analysis on the control instruction sequence, calculate the stress transfer chain and the deformation diffusion field between the solder joints, construct a state propagation network reflecting the dynamic evolution law of the group according to the stress transfer chain and the deformation diffusion field, and generate a group control topology; Analyze the local stability of the solder joint group based on the group control topology, calculate the singular value distribution of the solder joint stress field, identify the stress field fracture risk points and deformation accumulation points according to the singular value distribution, and generate a coupling influence matrix; Perform structural decomposition on the coupling influence matrix, extract the key control modes of the solder joint group, calculate the structural stiffness contribution spectrum of the solder joint group according to the key control modes, group and sort the solder joint execution sequences based on the structural stiffness contribution spectrum, and generate a group collaboration plan; Perform grouped decoupling on the control instruction sequence according to the group collaboration plan, calculate the energy distribution ratio of each group of solder joints, perform parameter reconstruction based on the energy distribution ratio, and perform compensation optimization in combination with the structural strength constraint to generate a group optimization control strategy.

[0052] The following specifically describes the steps involved in the above embodiments: Obtaining the spatial distribution data of the solder joint group and the sheet metal stress field data is achieved through the vehicle body digital twin system and predictive finite element analysis. The spatial distribution data includes the three-dimensional coordinates of each solder joint and the geometric relationship of the sheet metal where it is located, which is obtained by extracting vehicle body CAD data or using a three-dimensional coordinate measuring instrument, with the accuracy controlled within ±0.1 mm. The sheet metal stress field data is pre-calculated by finite element analysis software or obtained through actual measurement using strain gauges. Analyze the stress field of the control instruction sequence, regard the welding process of each solder joint as a local heat source, and calculate the temperature field and stress field distributions caused by welding through a thermal-structural coupling algorithm. Calculate the stress transfer chain between solder joints, that is, how stress is transferred from one solder joint to adjacent solder joints to form a stress transfer path. For example, at the connection part between the inner panel and the reinforcement panel of the door, due to the differences in material thickness and stiffness, an obvious stress transfer chain is formed, transferring from the thick plate area to the thin plate area, with a strength ratio of 3:1. At the same time, calculate the deformation diffusion field, which represents how the local deformation caused by welding spreads to the surrounding areas. Construct a state propagation network based on the stress transfer chain and the deformation diffusion field. This network uses solder joints as nodes and stress transfer paths as edges, and the edge weight represents the stress transfer strength. For example, in the network in the front area of the vehicle body side panel, the stress transfer strength between adjacent solder joints is usually between 65 - 120 MPa, while the transfer strength between distant solder joints is less than 30 MPa. This network modeling method based on physical mechanisms accurately captures the dynamic interaction relationship of the solder joint group as an overall structure.

[0053] Analyze the local stability of the solder joint group based on group control topology, and use the stability criterion in structural mechanics to evaluate the anti-deformation ability of the solder joint group under external forces. Calculate the singular value distribution of the solder joint stress field. The specific method is to perform singular value decomposition (SVD) on the incidence matrix of the state propagation network. The size of the singular value reflects the main mode of the stress distribution. The singular value decomposition result usually contains 10 - 15 main singular values, and the first 3 - 5 singular values contribute more than 80% of the total stress distribution. Identify two types of key points based on the singular value distribution: stress field fracture risk points, that is, areas where the singular value ratio (the ratio of the largest singular value to the second largest singular value) is greater than 5, indicating a high degree of stress concentration; deformation accumulation points, that is, positions where multiple singular vectors are superimposed and enhanced in the same area, indicating the combined action of multiple deformation modes. For example, in the connection area between the B-pillar and the sill of the vehicle body, the singular value ratio reaches 7.3, and a stress concentration area is identified, which needs to be controlled preferentially; while in the roof welding area, the deformation mode superposition coefficient reaches 2.8, forming a deformation accumulation area. Organize these analysis results into a coupling influence matrix between solder joints, and the matrix element Mij represents the influence strength of solder joint i on solder joint j. This method based on singular value analysis can identify the weak links and key control points in the structure starting from the overall structural characteristics of the solder joint group.

[0054] Perform a structural decomposition on the coupling influence matrix and use the matrix eigenvalue decomposition method to extract the key control modes of the solder joint group. The control mode refers to the characteristic deformation mode of the solder joint group as an overall structure, reflecting the intrinsic characteristics of the structure. By solving the characteristic equation \((A - \lambda I)v = 0\), the eigenvalues \(\lambda\) and eigenvectors \(v\) of the influence matrix \(A\) are obtained. The magnitude of the eigenvalue represents the importance of the mode, and the component of the eigenvector represents the participation degree of each solder joint in this mode. For the welding of the body side panel, the first 5 main modes usually explain more than 85% of the overall behavior. Calculate the structural stiffness contribution spectrum of the solder joint group according to the key control modes, that is, the contribution degree of each solder joint to the overall structural stiffness. The calculation method is to project the eigenvectors of each mode according to the solder joint position to obtain the stiffness contribution coefficient. For example, the stiffness contribution coefficient of a solder joint located at a key node of the stress transfer path is usually higher than 0.6, while that of a solder joint in a non-critical area is lower than 0.3. Group and sort the solder joint execution sequences based on the structural stiffness contribution spectrum, group the solder joints with similar contribution coefficients and adjacent spatial positions into the same group, and sort them from high to low according to the contribution degree. Typical body panel welding will form 3 - 5 solder joint groups, and each group contains 4 - 8 solder joints. This grouping method based on modal analysis realizes the scientific classification of complex solder joint groups and provides a theoretical basis for differential control strategies.

[0055] Group and decouple the control instruction sequence according to the group cooperation scheme, and reorganize the original control instructions for individual solder joints into a cooperative control strategy for solder joint groups. Calculate the energy distribution ratio of each group of solder joints, considering the structural importance and material properties of the solder joints. The calculation formula is E_i = E_base × W_i × M_i, where E_i is the energy distribution value of the i-th group of solder joints, E_base is the reference energy value, W_i is the structural stiffness contribution weight (0.6 - 1.2), and M_i is the material adjustment coefficient (0.8 - 1.5). For example, for the lap welding of high-strength steel and ordinary steel, the energy distribution of the solder joint group with a large contribution weight is increased by 15% to ensure the structural strength; while for the solder joint group in the material heterogeneous area, the energy is finely regulated within the range of ±8% of the reference value to avoid the weld nugget being too large or too small. Perform parameter reconstruction based on the energy distribution ratio and convert it into specific current waveforms, pressure curves, and timing control parameters. Combine with the structural strength constraint for compensation optimization to ensure that while meeting the single-point quality requirements, the overall structural performance is optimized. For example, for the solder joint group in the high-strength steel area of the vehicle body side panel, on the premise that the tensile strength of each solder joint is greater than 320 MPa, by adjusting the welding sequence and parameter distribution, the torsional stiffness of the overall structure is increased by 12% and the thermal deformation amount is reduced by 18%. The finally generated group optimization control strategy includes the complete solder joint execution sequence, group parameter configuration, and real-time compensation rules, forming a cooperative control scheme for the overall characteristics of the solder joint group. This group cooperation control method breaks through the limitations of traditional single-point control, optimizes the welding process from the overall structural performance, and significantly improves the welding quality and consistency of complex vehicle body structures.

[0056] In an embodiment of the present invention, the structural decomposition of the coupling influence matrix is performed to extract the key control modes of the solder joint group, the structural stiffness contribution spectrum of the solder joint group is calculated according to the key control modes, and the solder joint execution sequence is grouped and sorted based on the structural stiffness contribution spectrum to generate a group cooperation scheme, including: Perform singular value decomposition on the coupling influence matrix, extract the eigenvectors corresponding to the main singular values, construct a modal importance index according to the energy distribution characteristics of the eigenvectors, and generate a modal pedigree; Calculate the modal coupling strength between solder joints based on the modal pedigree, perform hierarchical analysis on the modal coupling strength, identify the key control modes according to the results of the hierarchical analysis, and generate a modal control sequence; Perform stress transfer analysis on the modal control sequence, calculate the contribution rate of each mode to the structural stiffness, construct a stiffness distribution function based on the contribution rate, and generate a stiffness contribution spectrum; Perform hierarchical clustering on the solder joint group according to the stiffness contribution spectrum, calculate the execution priority of each category of solder joints, design the solder joint execution sequence based on the execution priority, and generate a group cooperation scheme.

[0057] The following specifically describes the steps involved in the above embodiments: The singular value decomposition of the coupling influence matrix is implemented through a linear algebra calculation library. The n×n dimensional coupling influence matrix M is decomposed into M = UΣ , where U and V are orthogonal matrices, Σ is a diagonal matrix, and the diagonal elements are singular values σ1≥σ2≥...≥σₙ≥0. The calculation process uses an iterative numerical method, such as the Jacobi iterative algorithm, and the convergence accuracy is set to 10⁻ 6 . Extract the eigenvectors corresponding to the main singular values. Usually, the first k singular values with a cumulative contribution rate of more than 85% are selected. For typical body welding applications, the value of k is 4 - 7. Each component of the eigenvector represents the participation degree of the corresponding solder joint in this mode. Based on the energy distribution characteristics of the eigenvector, a modal importance index is constructed. The calculation method is the product of the L2 norm of the eigenvector and the corresponding singular value, that is, I_i = σ_i × ||v_i||2. For example, in the connection area between the B-pillar and the sill, the importance index of the first mode is 48.7, much higher than 23.5 of the second mode, indicating that the first mode is the dominant mode. Sort all modes from high to low according to their importance to generate a modal spectrum. This modal analysis method based on singular value decomposition reveals the main deformation modes of the solder joint group as a whole structure from a mathematical perspective, providing a theoretical basis for subsequent group collaborative control.

[0058] Calculate the modal coupling strength between solder joints based on the modal spectrum, using modal projection and correlation analysis methods. For each pair of solder joints (i,j), calculate the sum of the projection products on each mode to obtain the modal coupling coefficient C_ij. For example, for two solder joints on the outer panel of the car door with a distance of 50mm, the modal coupling coefficient is 0.78, indicating that these two solder joints are highly correlated in deformation behavior. Conduct hierarchical analysis on the modal coupling strength, using the hierarchical clustering algorithm, and gradually merge from the solder joint pair level to the solder joint group level. The complete linkage method is used in the clustering process, and the distance threshold is set to 0.25. This threshold is determined based on the body welding quality assessment experiment. Identify the key control modes according to the hierarchical analysis results, that is, the modal combinations that have the greatest impact on the overall structure deformation. For example, in the connection area between the front longitudinal beam and the shock absorber seat, the combination of the first and third modes explains 75% of the structural deformation. Sort these key control modes according to the degree of influence to generate a modal control sequence. This analysis method based on modal coupling reveals the functional association between solder joints, overcomes the limitation of the traditional method of grouping solder joints only based on spatial distance, and can identify solder joint groups that are closely related in structural function.

[0059] Perform stress transfer analysis on the modal control sequence using the stress path tracking algorithm. Apply a unit modal load in the finite element model to calculate the stress transfer path and intensity between solder joints. The identification of the stress transfer path uses the principal stress direction tracking method, and the stress intensity threshold is set to 15 MPa. Calculate the contribution rate of each mode to the structural stiffness, and the calculation method is the ratio of modal strain energy to total strain energy. For example, in the area of the front windshield pillar, the contribution rate of the first mode to the overall structural stiffness is 37%, and the second mode is 23%. Based on these contribution rates, construct the stiffness distribution function D(x, y, z), which describes the spatial distribution law of the structural stiffness. The stiffness distribution function is constructed using the radial basis function (RBF) interpolation method, with the Gaussian kernel selected as the kernel function and the smoothing coefficient set to 0.75. The generated stiffness contribution spectrum intuitively shows the degree of influence of each solder joint on the overall structural performance. For example, the stiffness contribution value of the solder joints located in the stress concentration area is usually in the range of 0.7 - 0.9, while the contribution value of the solder joints located in the low-stress area is less than 0.4. This stiffness analysis method based on stress transfer determines the structural importance of solder joints from the mechanical mechanism and provides a scientific basis for solder joint grading.

[0060] Classify and cluster the solder joint groups according to the stiffness contribution spectrum using the density-based spatial clustering algorithm DBSCAN. The algorithm parameters are set as follows: the neighborhood radius ε = 25 mm (determined based on the body panel spacing), and the minimum number of samples MinPts = 3 (ensuring that each group contains at least 3 solder joints). The clustering results divide the solder joints into 3 - 5 grades, and each grade represents different structural importance. Calculate the execution priority of each type of solder joint, and the calculation formula is P_i = 0.5×S_i + 0.3×R_i + 0.2×T_i, where S_i is the stiffness contribution value, R_i is the residual stress sensitivity, and T_i is the thermal deformation influence coefficient. The weight ratio of these three items (0.5:0.3:0.2) is determined based on the experiment on the relationship between body welding quality and structural performance. For example, for the welding of the outer panel of the side body, the execution priority of the first-grade solder joints is 0.85 and needs to be executed first; the execution priority of the third-grade solder joints is 0.42 and can be executed later. Design the execution sequence of solder joints based on the execution priority, and at the same time consider the optimization of the robot movement path to reduce the ineffective movement time. The finally generated group cooperation plan includes complete solder joint grouping information and an execution time sequence table. This solder joint sorting method based on structural importance ensures that key solder joints are given priority during the multi-solder joint cooperation process, significantly improving the stability of the overall welding quality and the consistency of the structural performance.

[0061] Please continue to refer to Figure 1, execute the population optimization control strategy, collect real-time control feedback data, calculate the real-time control deviation, and perform online correction on the parameter partition data, the initial control parameter set, the control instruction sequence, and the population optimization control strategy according to the real-time control deviation, so as to realize the adaptive adjustment of the control parameters.

[0062] In an embodiment of the present invention, the execution of the population optimization control strategy, the collection of real-time control feedback data, the calculation of the real-time control deviation, and the online correction of the parameter partition data, the initial control parameter set, the control instruction sequence, and the population optimization control strategy according to the real-time control deviation to realize the adaptive adjustment of the control parameters include: Execute the population optimization control strategy, collect the real-time data of welding current, electrode pressure, and displacement, extract the quality characteristics of the real-time data, calculate the nugget size deviation and the solder joint strength distribution, and generate a control deviation vector; Perform causal decomposition on the control deviation vector, identify the deviation components caused by process parameters, material state, and environmental factors, calculate the influence weights according to the transmission paths of the deviation components, and generate a deviation compensation sequence; Calculate the drift amount of the parameter partition boundary based on the deviation compensation sequence, adjust the boundaries of the parameter partition data and the initial control parameter set, correct the parameter mapping relationship according to the boundary adjustment amount, and generate a corrected parameter set; Map the corrected parameter set into the control instruction sequence and the population optimization control strategy, perform gradient update on the control parameters according to process constraints, and perform compensation correction according to welding quality constraints to realize the adaptive adjustment of the control parameters.

[0063] The following specifically describes the steps involved in the above embodiments: Implementing the swarm optimization control strategy and collecting real-time data during the welding process need to be completed through the welding quality monitoring system. This system consists of high-precision data acquisition modules, including a Hall current sensor (sampling rate 20 kHz, accuracy ±0.5%) to collect the welding current waveform, a pressure sensor (response time <1 ms) to monitor the electrode pressure change, and a laser displacement sensor (resolution 0.1 μm) to measure the electrode displacement. The collected real-time data, after signal conditioning, enters the quality feature extraction module, which uses dynamic feature analysis algorithms to extract key indicators reflecting the welding quality. First, the nugget size is calculated based on the electrode displacement curve and the resistance change rate, and the modified Chertov model is used to estimate the nugget diameter. Then, the actual nugget size is compared with the expected target value (determined by the process requirements) to calculate the nugget size deviation. At the same time, the solder joint strength distribution is analyzed by combining the dynamic resistance and energy input, and a neural network model is used to map the resistance curve features to strength estimates. For example, in the welding of high-strength steel for the body B-pillar, when the actual nugget diameter is 4.8 mm and the target value is 5.2 mm, the size deviation is -7.7%; the strength distribution shows an uneven characteristic with higher strength in the center (about 420 MPa) and lower strength at the edge (about 345 MPa). These deviation data form a multi-dimensional control deviation vector, accurately reflecting the difference between the welding quality and the target. This quality monitoring method based on multi-sensor fusion breaks through the limitations of traditional single-index evaluation and can comprehensively capture the multi-dimensional characteristics of welding quality.

[0064] Causal decomposition of the control deviation vector is achieved through the structural equation model (SEM). First, a causal network model including process parameters, material state, and environmental factors is established. Then, through the Bayesian inference method, the observed control deviation vector is decomposed into deviation components caused by different factors. The process parameter deviation components mainly include current waveform error, pressure control deviation, etc.; the material state deviation components include surface state fluctuations, material thickness changes, etc.; the environmental factor deviation components include electrode wear, heat dissipation condition changes, etc. For example, in the welding of the outer panel of a car door, the problem of a smaller nugget size may be analyzed to consist of 35% contribution from process parameters (too low current waveform rise rate), 45% contribution from material state (insufficient surface cleanliness), and 20% contribution from environmental factors (electrode wear). The influence weights are calculated based on the transfer paths of these deviation components, and the transfer paths are traced through the gradient propagation algorithm to find the influence chain of the deviation from the root cause to the final quality feature. The influence weight calculation takes into account three factors: path length, transfer strength, and time delay, and the weight allocation ratio is 3:5:2. The generated deviation compensation sequence includes the compensation amount and compensation timing for each type of deviation. This deviation decomposition method based on causal analysis accurately identifies the root cause of quality problems from the physical mechanism, avoiding blind compensation.

[0065] The drift of the parameter partition boundary is calculated based on the deviation compensation sequence and implemented using the boundary drift tracking algorithm. This algorithm identifies systematic drifts in the parameter space by analyzing the adjustment trends of the parameters in the deviation compensation sequence. The drift calculation takes into account short-term fluctuations and long-term trends, using the exponentially weighted moving average method with a short-term weight of 0.3 and a long-term weight of 0.7. The boundaries of the parameter partition data and the initial control parameter set are adjusted, and the adjustment step size is adaptively set according to the severity of the drift, generally controlled within the range of 5 - 15% of the original boundary value. For example, when the electrode wear causes a systematic increase in the contact resistance, the partition boundary of the heat input intensity parameter needs to be adjusted upward by about 10% to compensate for the increased heat loss. The parameter mapping relationship is corrected according to the boundary adjustment amount, that is, the conversion function connecting the original parameter space and the actual process space is updated. The mapping relationship correction uses the piecewise linear interpolation method to ensure the continuity of the parameter space while adjusting the boundary. The generated corrected parameter set contains the updated parameter boundary values and the corresponding mapping function parameters. This dynamic boundary adjustment method based on drift tracking enables the control system to adapt to slow-changing factors in the production process, such as electrode wear and equipment aging, and maintain long-term stable control effects.

[0066] The corrected parameter set is mapped into the control instruction sequence and the population optimization control strategy through the parameter remapping algorithm. This algorithm first establishes a correspondence table between the corrected parameters and the control instructions, and then updates the specific values of the control instructions according to the mapping rules. The control parameters are updated in gradients according to the process constraints, and the update step size adopts an adaptive method. For important parameters (parameters that affect the welding quality > 30%), a small step size (3 - 5%) is used, for secondary parameters, a medium step size (5 - 10%) is used, and for compensation parameters, a large step size (10 - 15%) is used. The process constraints are strictly followed during the update process, such as the current rise rate not exceeding 200 A / ms (to prevent spatter) and the pressure change rate not exceeding 3 kN / ms (to protect the electrode). For example, in the welding of the front windshield frame, when the surface zinc layer fluctuation is detected, the system increases the current in the preheating stage by 7%, while keeping the current in the main heating stage unchanged, and slightly extends the pressure curve by 15 ms. Compensation correction is performed according to the welding quality constraints to ensure that the adjusted parameters still meet the strength requirements and appearance requirements. The correction process adopts a closed-loop control strategy, and the effect is evaluated after each adjustment, and continuous adjustment is performed until the deviation is less than the threshold (usually set to 5%) or the maximum number of iterations is reached (generally 5 times). This adaptive adjustment method based on multiple constraints realizes the intelligent optimization of welding parameters, maximally adapts to the dynamic changes of the production environment while ensuring the welding quality, and improves the stability and reliability of the welding process.

[0067] The above describes the body flexible welding process in the embodiments of the present invention. Next, the body flexible welding device in the embodiments of the present invention will be described. Please refer to Figure 2, an embodiment of the body flexible welding device in the embodiments of the present invention includes: A parameter partitioning module 101, configured to obtain multi-dimensional parameter data and perform partitioning processing to obtain core quality area parameters, stable equilibrium area parameters, and environmental adaptation area parameters, calculate the influence weights of each parameter on the process respectively, determine the control priority based on the influence weights, and generate a resource allocation scheme; A working condition mapping module 102, configured to convert target working condition data into a working condition feature vector, calculate the similarity with historical working condition data according to the working condition feature vector, and output initial control parameters based on the similarity and the resource allocation scheme; A timing control module 103, configured to obtain process state data based on the initial control parameters, calculate the current change rate and temperature gradient according to the state data, determine the process stage, perform microsecond-level control parameter adjustment on the process stage, generate a dynamic intervention instruction, execute parameter regulation according to the dynamic intervention instruction, and output optimized control parameters; A group cooperation module 104, configured to obtain multi-point distribution data, calculate the influence relationship according to the multi-point distribution data, construct a network topology structure, analyze the temperature field and stress field distributions based on the network topology structure, generate a regional balance compensation scheme, and perform compensation adjustment on the optimized control parameters according to the regional balance compensation scheme, and output global coordination parameters; A material adaptation module 105, configured to establish a control mode space according to material characteristic data, generate a parameter gradual change sequence based on the control mode space, and perform continuous adjustment on the global coordination parameters according to the parameter gradual change sequence, and output final control parameters.

[0068] Above Figure 2 The medium body flexible welding device in the embodiments of the present invention is described in detail from the perspective of modular functional entities. Below, the body flexible welding equipment in the embodiments of the present invention is described in detail from the perspective of hardware processing.

[0069] Figure 3It is a schematic structural diagram of a body flexible welding device provided by an embodiment of the present invention. The body flexible welding device 200 can vary greatly due to different configurations or performances, and may include one or more processors (central processing units, CPU) 210 (for example, one or more processors) and a memory 220, and one or more storage media 230 (for example, one or more mass storage device terminals) for storing application programs 233 or data 232. Among them, the memory 220 and the storage media 230 can be transient storage or persistent storage. The program stored in the storage media 230 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the body flexible welding device 200. Further, the processor 210 can be set to communicate with the storage media 230 and execute a series of instruction operations in the storage media 230 on the body flexible welding device 200 to implement the steps of the above-mentioned body flexible welding process.

[0070] The body flexible welding device 200 may further include one or more power supplies 240, one or more wired or wireless network interfaces 250, one or more input / output interfaces 260, and / or one or more operating systems 231, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, and so on. Those skilled in the art can understand that Figure 3 The shown structural diagram of the body flexible welding device does not constitute a limitation on the body flexible welding device provided by the present invention, and may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.

[0071] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions run on a computer, the computer is made to execute the steps of the body flexible welding process.

[0072] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described system or device and unit can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0073] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0074] The foregoing are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structural transformation made by using the content of the specification and drawings of the present invention under the inventive concept of the present invention, or any direct / indirect application in other related technical fields is included in the patent protection scope of the present invention.

Claims

1. A flexible body welding process, characterized in that, Including: Obtain system operation parameter data, calculate the change rate and response characteristics of the operation parameters, perform real-time partitioning of the parameter space according to the change rate and the response characteristics, divide the parameter space into a core control parameter group, a steady-state balance parameter group, and a compensation adjustment parameter group, and generate parameter partitioning data; Extract the system state feature vector according to the core control parameter group in the parameter partitioning data, calculate the deviation value between the state feature vector and the preset control target, retrieve a similar control strategy from the control experience library based on the deviation value, and generate an initial control parameter group; Collect system real-time feedback data, calculate the system dynamic response characteristics based on the initial control parameter group, identify the key control moments of the system response according to the dynamic response characteristics, dynamically correct the initial control parameter group at the key control moments, and output a control instruction sequence; Input the control instruction sequence into the group control network, calculate the coupling influence matrix between the control objects, and perform collaborative optimization on the control instruction sequence according to the coupling influence matrix to generate a group optimization control strategy; Execute the group optimization control strategy and collect real-time control feedback data, calculate the real-time control deviation, and perform online correction on the parameter partitioning data, the initial control parameter group, the control instruction sequence, and the group optimization control strategy according to the real-time control deviation to achieve adaptive adjustment of the control parameters.

2. The body flexible welding process according to claim 1, characterized in that, The obtaining system operation parameter data, calculating the change rate and response characteristics of the operation parameters, performing real-time partitioning of the parameter space according to the change rate and the response characteristics, dividing the parameter space into a core control parameter group, a steady-state balance parameter group, and a compensation adjustment parameter group, and generating parameter partitioning data includes: Obtain the operation parameter data of the heat input intensity, electrode pressure, and position compensation amount, calculate the parameter change rate according to the operation parameter data, perform multi-point sampling on the parameter change rate, and generate a parameter dynamic sequence; Calculate the ratio of the parameter response time to the material softening time based on the parameter dynamic sequence, use the ratio as the parameter time series characteristic, construct a parameter response function according to the parameter time series characteristic, and generate a parameter response characteristic; Perform cross-mapping analysis on the parameter dynamic sequence and the parameter response characteristic, calculate the causal association strength between the parameters, construct a parameter link network based on the causal association strength, and generate a parameter coupling relationship; Identify the key nodes of the parameter control chain based on the parameter coupling relationship, calculate the fan-in coefficient and fan-out coefficient of the key nodes, evaluate the parameter influence based on the fan-in coefficient and the fan-out coefficient, and generate the parameter importance; Classify the parameters according to the parameter importance, assign the parameters with an influence higher than the first threshold to the core control parameter group, assign the parameters with an influence between the first threshold and the second threshold to the steady-state balance parameter group, and assign the parameters with an influence lower than the second threshold to the compensation adjustment parameter group to generate parameter partitioning data.

3. The body flexible welding process according to claim 1, characterized in that, Extracting a system state feature vector according to the core control parameter group in the parameter partition data, calculating a deviation value between the state feature vector and a preset control target, and retrieving a similar control strategy from a control experience library based on the deviation value to generate an initial control parameter group, including: Extracting welding current density and pressure distribution data according to the core control parameter group in the parameter partition data, calculating a predicted value of the nugget size according to the welding current density and the pressure distribution data, and constructing a nugget growth feature in combination with the solder joint position information to generate a state feature component; Obtaining a resistance change curve between the electrode and the workpiece, performing segmented processing on the resistance change curve, and calculating a slope sequence of the resistance change curve in different welding stages to generate a dynamic response component; Orthogonally combining the state feature component and the dynamic response component to construct a state feature vector reflecting the physical essence of the welding process, and calculating a deviation value of the state feature vector in a multi-dimensional process space according to a preset welding quality target; Establishing a welding condition similarity evaluation function according to the deviation value, retrieving a similar welding condition from a control experience library based on the welding condition similarity evaluation function, and extracting parameter configurations corresponding to the similar welding condition to generate a candidate control strategy; Performing process window verification on the candidate control strategy, calculating the process feasibility of the candidate control strategy, and screening parameter configurations that meet process constraints according to the process feasibility to generate an initial control parameter group.

4. The body flexible welding process according to claim 1, characterized in that, The acquisition system real-time feedback data, calculating system dynamic response characteristics based on the initial control parameter group, identifying key control moments of system response according to the dynamic response characteristics, and dynamically correcting the initial control parameter group at the key control moments to output a control instruction sequence, including: Collecting real-time feedback data of the current waveform, pressure waveform, and electrode displacement amount during the welding process, performing segmented response analysis on the current parameters and pressure parameters in the initial control parameter group, and calculating the energy input efficiency and heat loss ratio of each segment to generate a multi-segment response feature; Calculating the energy balance state of the welding process according to the multi-segment response feature, extracting energy mutation points and thermal equilibrium points during the welding process, and constructing a state transition diagram of the welding process based on the energy mutation points and the thermal equilibrium points to generate a process state sequence; Performing energy gradient analysis on the process state sequence, identifying the demarcation points of the nugget formation period, transition period, and stable period, and calculating the critical energy threshold for state conversion according to the demarcation points to generate key control moments; Establishing a parameter compensation mapping function based on the key control moments and the critical energy threshold, mapping the initial control parameter group to the parameter space of different welding stages, and calculating a parameter adjustment amount according to the critical energy threshold to generate segmented control parameters; Performing energy continuity verification on the segmented control parameters, calculating the energy jump amount during parameter switching, and performing parameter smoothing compensation according to the energy jump amount to output a control instruction sequence.

5. The body flexible welding process according to claim 4, characterized in that, Performing an energy gradient analysis on the process state sequence to identify the demarcation points of the nugget formation period, transition period, and stable period, and calculating the critical energy threshold for state transition according to the demarcation points to generate key control moments, including: Calculating the energy accumulation curve of the process state sequence, performing a differential operation on the energy accumulation curve to obtain an energy change rate curve, constructing an energy fluctuation index according to the fluctuation characteristics of the energy change rate curve, and generating an energy characteristic sequence; Performing a wavelet transform on the energy characteristic sequence, extracting energy mutation characteristics and energy gradual change characteristics, identifying energy conversion points according to the combined pattern of the energy mutation characteristics and the energy gradual change characteristics, and generating a state transition sequence; Calculating the energy difference between adjacent conversion points according to the state transition sequence, performing a clustering analysis on the energy difference, and identifying the energy transition characteristics of the nugget formation period, transition period, and stable period based on the clustering analysis result to generate a demarcation point sequence; Performing an energy density analysis on the demarcation point sequence, calculating the energy density gradient at each demarcation point, and determining the critical energy threshold for state transition according to the energy density gradient to generate key control moments.

6. The body flexible welding process according to claim 1, wherein, Inputting the control instruction sequence into a group control network, calculating the coupling influence matrix between control objects, and performing collaborative optimization on the control instruction sequence according to the coupling influence matrix to generate a group optimization control strategy, including: Obtaining the spatial distribution data of the solder joint group and the sheet metal stress field data, performing a stress field analysis on the control instruction sequence, calculating the stress transfer chain and deformation diffusion field between solder joints, and constructing a state propagation network reflecting the group dynamic evolution law according to the stress transfer chain and the deformation diffusion field to generate a group control topology; Analyzing the local stability of the solder joint group based on the group control topology, calculating the singular value distribution of the solder joint stress field, and identifying the stress field fracture risk points and deformation accumulation points according to the singular value distribution to generate a coupling influence matrix; Performing a structure decomposition on the coupling influence matrix, extracting the key control modes of the solder joint group, calculating the structure stiffness contribution spectrum of the solder joint group according to the key control modes, and performing a grouping and sorting on the solder joint execution sequence based on the structure stiffness contribution spectrum to generate a group collaboration plan; Performing a grouping and decoupling on the control instruction sequence according to the group collaboration plan, calculating the energy distribution ratio of each group of solder joints, performing a parameter reconstruction based on the energy distribution ratio, and performing a compensation optimization in combination with the structural strength constraint to generate a group optimization control strategy.

7. The body flexible welding process according to claim 6, characterized in that, Performing a structure decomposition on the coupling influence matrix, extracting the key control modes of the solder joint group, calculating the structure stiffness contribution spectrum of the solder joint group according to the key control modes, and performing a grouping and sorting on the solder joint execution sequence based on the structure stiffness contribution spectrum to generate a group collaboration plan, including: Performing a singular value decomposition on the coupling influence matrix, extracting the eigenvectors corresponding to the main singular values, constructing a modal importance index according to the energy distribution characteristics of the eigenvectors, and generating a modal spectrum; Calculate the modal coupling strength between solder joints based on the modal pedigree, perform hierarchical analysis on the modal coupling strength, identify key control modes according to the results of the hierarchical analysis, and generate a modal control sequence; Perform stress transfer analysis on the modal control sequence, calculate the contribution rate of each mode to the structural stiffness, construct a stiffness distribution function based on the contribution rate, and generate a stiffness contribution spectrum; Perform hierarchical clustering on the solder joint group according to the stiffness contribution spectrum, calculate the execution priority of each type of solder joint, design the execution order of solder joints based on the execution priority, and generate a group cooperation plan.

8. The body flexible welding process according to claim 1, characterized in that, Execute the group optimization control strategy and collect real-time control feedback data, calculate the real-time control deviation, and perform online correction on the parameter partition data, the initial control parameter group, the control instruction sequence, and the group optimization control strategy according to the real-time control deviation to achieve adaptive adjustment of control parameters, including: Execute the group optimization control strategy and collect real-time data of welding current, electrode pressure, and displacement, extract quality characteristics from the real-time data, calculate the nugget size deviation and the solder joint strength distribution, and generate a control deviation vector; Perform causal decomposition on the control deviation vector, identify the deviation components caused by process parameters, material states, and environmental factors, calculate the influence weights according to the transfer paths of the deviation components, and generate a deviation compensation sequence; Calculate the drift amount of the parameter partition boundary based on the deviation compensation sequence, perform boundary adjustment on the parameter partition data and the initial control parameter group, and correct the parameter mapping relationship according to the boundary adjustment amount to generate a corrected parameter group; Map the corrected parameter group into the control instruction sequence and the group optimization control strategy, perform gradient update on the control parameters according to process constraints, and perform compensation correction according to welding quality constraints to achieve adaptive adjustment of control parameters.

9. A flexible vehicle body welding device, characterized in that, The body flexible welding device adopts the body flexible welding process according to any one of claims 1 to 8, and the body flexible welding device includes: A parameter partition module, configured to obtain multi-dimensional parameter data and perform partition processing to obtain core quality area parameters, stable equilibrium area parameters, and environmental adaptation area parameters, calculate the influence weights of each parameter on the process respectively, determine the control priority based on the influence weights, and generate a resource allocation plan; A working condition mapping module, configured to convert target working condition data into a working condition feature vector, calculate the similarity with historical working condition data according to the working condition feature vector, and output initial control parameters based on the similarity and the resource allocation plan; A timing control module, configured to obtain process state data based on the initial control parameters, calculate the current change rate and the temperature gradient according to the state data, determine the process stage, perform microsecond-level control parameter adjustment on the process stage, generate a dynamic intervention instruction, perform parameter regulation according to the dynamic intervention instruction, and output optimized control parameters; The group collaboration module is used to obtain multi-point distribution data, calculate influence relationships according to the multi-point distribution data, construct a network topology structure, analyze the temperature field and stress field distributions based on the network topology structure, generate a regional balance compensation plan, and compensate and adjust the optimized control parameters according to the regional balance compensation plan, and output global coordination parameters; The material adaptation module is used to establish a control mode space according to material characteristic data, generate a parameter gradual change sequence based on the control mode space, continuously adjust the global coordination parameters according to the parameter gradual change sequence, and output final control parameters.

10. A flexible body welding device, characterized in that, The body flexible welding device includes: a memory and at least one processor, and instructions are stored in the memory; The at least one processor calls the instructions in the memory so that the body flexible welding device executes the steps of the body flexible welding process according to any one of claims 1 to 8.

Images