Method and device for regulating and controlling welding parameters of composite material

Abstract

The invention discloses a composite material welding parameter regulation and control method and device, and belongs to the technical field of intelligent welding, and the method comprises the steps of constructing a process coupling model, setting a critical positioning method, generating a parameter bifurcation atlas, constructing a chaotic boundary probability cloud atlas, associating image feature differences of different materials, and positioning a parameter critical region; obtaining a dynamic data set, setting a critical early warning method, extracting key features, analyzing abnormal changes of image features, and triggering early warning once the abnormal changes are recognized; calculating a safety offset, setting a dynamic reconstruction method, introducing a time-varying damping factor, dynamically adjusting collaborative output, and reconstructing an optimal parameter combination in combination with a welding depth and an interface fusion state fed back by a real-time image; the method comprises the steps of collecting batch welding data, setting an iterative optimization method, updating process coupling model parameters, dynamically expanding a critical region recognition range, correcting bifurcation boundary conditions in real time and enhancing the adaptability to parameter critical point drift.

Description

TECHNICAL FIELD

[0001] The application relates to a composite material welding parameter regulation method and device, and belongs to the technical field of intelligent welding. BACKGROUND

[0002] Ultrasonic welding realizes molecular layer bonding by high-frequency mechanical vibration to generate friction heat on the material contact surface, and the welding quality is closely related to welding parameters.

[0003] However, the prior art does not consider the critical point drift problem of the nonlinear process, especially the welding quality risk caused by the chaos effect when the welding parameters approach the critical value of the nonlinear dynamics; specifically, in the ultrasonic welding process implemented by a mechanical arm, the movement control of the mechanical arm and the pressure, ultrasonic frequency, welding depth and ultrasonic amplitude adjustment constitute a nonlinear welding process mode, when the related parameters approach the critical threshold, the chaos state is easily triggered, for example, when the ultrasonic frequency approaches the resonance critical point of the combination of the transducer, the amplitude rod and the welding head, heat energy is generated in the actual working process, which reduces the amplitude and affects the welding strength, and the mechanical properties of the product are reduced. In addition, the slight change in the stiffness of the welding head caused by the change in the load of the mechanical arm will be amplified through nonlinear acoustic coupling, which will cause the actual vibration frequency to break through the critical value, and then cause unpredictable violent fluctuations in key welding parameters such as heat input. This abnormal fluctuation of the parameters causes overheating carbonization or unmelting defects on the welding interface: the molecular chains of the material in the overheated area are broken, which greatly reduces the strength; the unmelting area forms a weak part with an interfacial peeling strength lower than the standard value, which affects the stability of the welding quality. SUMMARY

[0004] In view of the deficiencies of the prior art, the purpose of the present application is to provide a composite material welding parameter regulation method and device, which realizes accurate identification and dynamic regulation of the critical drift of the nonlinear process through deep fusion of data and image features, effectively suppresses heat input fluctuations, reduces overheating carbonization, unmelting and other defects, and improves the stability of the welding quality.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0006] A composite material welding parameter regulation method, comprising:

[0007] Constructing a process coupling model, setting a critical positioning method, generating a parameter bifurcation atlas, and constructing a chaos boundary probability cloud map, and associating the image feature differences of different materials to locate the parameter critical region;

[0008] Obtaining a multispectral sequence, setting a critical early warning method, extracting key features, and triggering a pre-alarm by analyzing the abnormal changes of the image features once the abnormal changes are identified;

[0009] The optimal parameter combination is reconstructed by calculating the safety offset, setting the dynamic reconstruction method, introducing the time-varying damping factor, dynamically adjusting the cooperative output, and combining the welding depth and the interface fusion state of the real-time image feedback.

[0010] The batch welding data is collected, the iterative optimization method is set, the process coupling model parameters are updated, the critical region identification range is dynamically expanded, and the bifurcation boundary conditions are corrected in real time.

[0011] Specifically, the critical positioning method comprises:

[0012] The historical parameter data under dynamic working conditions is collected and processed in layers, the image sequence is aligned between frames and filtered by Gauss, and a layered data set is generated;

[0013] Cross-scale image feature parameter extraction is performed, the surface visible light image of the material is super-resolution reconstructed, the surface texture entropy is extracted, the surface micro-defects are segmented and the defect density and distribution entropy are calculated, the temperature field inversion is performed on the infrared thermal imaging image based on the heat conduction equation, the highest temperature and temperature gradient slope of the interface are extracted, the feature parameters are clustered according to the material type, and a feature parameter set is generated;

[0014] Based on the ultrasonic vibration energy transmission equation and the composite material fusion dynamics model, the theoretical constraints between parameters are established, the cross-scale image feature parameters are introduced into the model as soft constraints, the process coupling model is constructed, and the historical welding data is used to adjust the weight coefficient of the image feature by using the Bayesian optimization algorithm to train the process coupling model;

[0015] The basic critical threshold is set, the image features and quality fluctuation data in the trial welding process are extracted, the sensitivity is calculated, and the basic critical threshold is dynamically corrected to generate a critical positioning rule.

[0016] Specifically, the critical positioning method further comprises:

[0017] A three-dimensional coordinate axis is constructed, a plurality of parameter combinations are generated within a preset parameter range, a bifurcation point is marked in combination with the process coupling model, and the bifurcation point is divided into a stable region, a subcritical region and a chaotic region to generate a parameter bifurcation atlas;

[0018] For the bifurcation point in the parameter bifurcation atlas, a plurality of perturbation parameters are generated, the reference probability distribution is calculated through the process coupling model, and a probability cloud map is drawn in the three-dimensional coordinate axis;

[0019] For new materials, the probability cloud map of similar materials is retrieved from the historical database, and the domain self-adaptive transfer learning algorithm is used to transfer the probability distribution of similar materials to new materials; wherein, the feature parameter set of the new material is extracted, the Euclidean distance calculation is performed between the feature vector of the material in the historical database, and the materials with a Euclidean distance less than a similarity difference threshold are defined as similar materials.

[0020] Collecting microscopic images of joint wear of a mechanical arm and calculating joint correlation coefficients to form a chaotic boundary probability cloud map by integrating a probability cloud map;

[0021] Based on the chaotic boundary probability cloud map, dividing into a low-risk area, a warning area, and a high-risk area according to chaotic probability values, and establishing a correlation matrix by statistically analyzing the distribution of image feature parameters in each area;

[0022] Calculating the deviation of the actual chaotic probability value from the predicted value, and adjusting the region boundary when the deviation is greater than the correction threshold to form a critical region parameter set.

[0023] Specifically, the critical warning method comprises:

[0024] Real-time image data is collected and cross-scale image feature parameters are extracted to generate a real-time feature sequence;

[0025] Based on the critical region parameter set, a stable critical threshold is obtained, and if any feature in the real-time feature sequence exceeds the stable critical threshold, it is marked as a potential abnormal feature, otherwise it is determined to be normal and continuous abnormality monitoring is performed;

[0026] For the potential abnormal feature, a single-feature deviation rate is calculated and weighted to obtain a comprehensive abnormality degree of the real-time feature sequence.

[0027] Specifically, the critical warning method further comprises:

[0028] The parameter combination is back calculated in combination with the process coupling model, and the region is located in the parameter bifurcation atlas, and the chaotic probability change rate is obtained by querying the chaotic boundary probability cloud map;

[0029] If the comprehensive abnormality degree is greater than the abnormality threshold and the chaotic probability change rate is greater than the abnormality rate threshold, it is determined to be critical drift;

[0030] The critical drift is classified and warned, and if the comprehensive abnormality degree is between the abnormality threshold and the secondary abnormality threshold and the parameter combination is located in the subcritical region, a first-level warning is triggered, and an abnormal feature time sequence curve is generated, otherwise a second-level warning is triggered.

[0031] Specifically, the dynamic reconstruction method comprises:

[0032] A second-level warning signal is received, the location of the warning parameter combination is located in the parameter bifurcation atlas, defined as a risk area, and the stable critical threshold of the risk area is obtained, and the basic offset is calculated;

[0033] In combination with the correlation matrix, a coupling coefficient is introduced to correct the basic offset, and a material correction coefficient is called to correct the basic offset again;

[0034] The drift amount of the real-time characteristic parameter is calculated, and the comprehensive drift amount is calculated by taking the sensitivity as a weight and combining the load drift amount;

[0035] The drift acceleration is obtained, the time-varying damping factor is calculated by combining the basic coefficient and the dynamic coefficient, and the initial pressure and the initial amplitude are cooperatively corrected;

[0036] The real-time welding depth is obtained, the target welding depth is called, the target welding deviation is calculated, and once the target welding deviation is greater than the deviation threshold, the control parameter is adjusted based on the PID algorithm to generate a compensation parameter set.

[0037] Specifically, the dynamic reconstruction method further comprises:

[0038] A rolling period is set, a dynamic target function is constructed, and a constraint condition is set;

[0039] A particle swarm algorithm with a mutation operator is used to search for an optimal parameter combination in a safety zone and to adjust the parameters;

[0040] The feature parameters of three frames of images after adjustment are collected, and the average deviation rate is calculated;

[0041] If the average deviation rate is less than the deviation verification threshold, it is determined that the adjustment is effective, otherwise it is determined that the adjustment is ineffective;

[0042] A plurality of rolling periods are continuously monitored, and the standard deviation is calculated, if the standard deviation is less than the standard verification threshold, it is determined to be out of risk, otherwise it is determined to be not out of risk;

[0043] If any verification fails, the parameter adjustment and compensation process is repeated until all indicators meet the stable state requirements, and the optimal parameter combination is generated.

[0044] Specifically, the iterative optimization method comprises:

[0045] Various key data in the welding process are summarized, and the key data are structured processed;

[0046] The real-time image features in the welding process and the interface image features after welding are time-aligned, the similarity is calculated, and the matching samples with a similarity exceeding a correlation threshold are screened out;

[0047] For the matching samples, the linear correlation degree and the nonlinear correlation degree of the image feature drift amount and the parameter drift amount are calculated, the quality influence weight is calculated, and a correlation rule library is generated;

[0048] The process coupling model is updated in layers, incremental learning is performed using a federated learning framework, sub-models are trained in blocks according to welding batches, sub-model parameters are fused by weighted averaging, a verification set is constructed, and an incrementally optimized process coupling model is output.

[0049] Specifically, the iterative optimization method further comprises:

[0050] Collecting relevant parameters of new materials, comparing with historical material characteristic library, and constructing characteristic difference vector;

[0051] If the length of the characteristic difference vector is greater than the filing threshold, it is determined as high difference material, a trial welding group is constructed, trial welding parameters are collected, and identification parameters are backstepped, and the process coupling model and the characteristic difference vector are updated based on the identification parameters to correct the bifurcation boundary;

[0052] Otherwise, it is determined as approximately matched material, and the initial scheme is called to fine-tune the bifurcation boundary.

[0053] A composite material welding parameter control device comprises a critical positioning module, a critical early warning module, a dynamic reconstruction module and an iterative optimization module.

[0054] The critical positioning module is used to construct a process coupling model, generate a parameter bifurcation atlas, and construct a chaotic boundary probability cloud map, and associate image characteristic differences of different materials to locate a parameter critical region.

[0055] The critical early warning module is used to acquire a multi-spectrum sequence, extract key features, and trigger a pre-alarm by analyzing abnormal changes of image features once abnormal changes are identified.

[0056] The dynamic reconstruction module is used to calculate a safety offset, dynamically adjust a cooperative output, and combine a welding depth and an interface fusion state fed back in real time to reconstruct an optimal parameter combination.

[0057] The iterative optimization module is used to collect batch welding data, update process coupling model parameters, dynamically expand a critical region identification range, and correct bifurcation boundary conditions in real time.

[0058] The beneficial effects of the present application are as follows:

[0059] Based on the coupling mechanism of mechanical arm motion, ultrasonic vibration and material response, a process coupling model is established combined with cross-scale image features. The parameter critical region is accurately located through the parameter bifurcation atlas and chaotic boundary probability cloud, which breaks through the limitation of traditional methods that are difficult to identify the drift of critical point. The abnormal fluctuation trend of parameters caused by the ultrasonic frequency close to the resonance critical point and the change of mechanical arm load can be captured in advance. With the help of high-resolution visual monitoring and multi-dimensional feature analysis, the nonlinear abnormalities of image features are captured in real time, the early warning of chaotic state is realized, and the problems such as amplitude drop and heat input fluctuation caused by the parameters breaking through the critical value are avoided. For the early warning signal, the safety offset is calculated and the time-varying damping factor is introduced through dynamic parameter reconstruction, and the pressure, amplitude and welding depth are adjusted coordinately to offset the nonlinear amplification effect, stabilize the heat input, and effectively inhibit the generation of overheated carbonization or un-melted defects. Through the whole process data iterative optimization, the model and the critical region boundary are dynamically updated by combining parameter drift, image features and quality feedback, the adaptability to the drift of critical point in the scene of new materials and equipment wear is enhanced, and finally the precise control of chaotic effect in nonlinear process is realized, which significantly improves the stability of welding quality and avoids the problems such as strength reduction and interface defects caused by abnormal fluctuation of parameters. BRIEF DESCRIPTION OF DRAWINGS

[0060] Figure 1 It is a schematic diagram of a composite material welding parameter control method.

[0061] Figure 2 It is a flow chart of the critical positioning method in the present application.

[0062] Figure 3 It is a flow chart of the critical early warning method in the present application.

[0063] Figure 4 It is a flow chart of the dynamic reconstruction method in the present application.

[0064] Figure 5 It is a flow chart of the iterative optimization method in the present application. DETAILED DESCRIPTION

[0065] The technical scheme of the present application will be described in detail below with the help of the drawings and specific embodiments. It should be understood that the specific features in the embodiments and examples of the present application are detailed descriptions of the technical scheme of the present application, and are not limitations of the technical scheme of the present application. In the case of no conflict, the technical features in the embodiments and examples of the present application can be combined with each other.

[0066] Embodiment 1:

[0067] Reference Figures 1 to 5 As shown in the figure, the present embodiment introduces a composite material welding parameter control method, which includes the following steps:

[0068] Step S1: Based on the mechanical arm motion characteristics, ultrasonic vibration transmission path and material response physical mechanism, combined with the image characteristics of the welding interface and the material surface, a multi-parameter coupled process coupling model is established, including pressure, welding depth, vibration frequency, ultrasonic amplitude, image feature parameters, a critical positioning method is set, a parameter bifurcation atlas is generated through simulation analysis, combined with the image feature differences of different materials, a chaos boundary probability cloud diagram of the combination of mechanical arm speed and vibration frequency is constructed, and the image feature differences of different materials are associated, the chaos boundary in the parameter space is determined, and the parameter critical region is located;

[0069] Step S2: Use high-resolution visual acquisition device to continuously acquire dynamic data set, including mechanical arm end welding head motion trajectory image, welding interface real-time state image and material surface micro feature image, set critical early warning method, combined with edge detection, texture entropy analysis, extract key features from image, through analyzing abnormal changes of image features, capture nonlinear features before parameters approach critical value, once abnormal changes are identified, trigger pre-alarm, realize early warning of chaotic state;

[0070] Step S3: After triggering the pre-alarm, based on the parameter bifurcation atlas, calculate the safety offset of the current parameter point to the subcritical region, synchronously adjust the mechanical arm motion speed and ultrasonic amplitude, migrate the parameter combination to the subcritical region, set dynamic reconstruction method, introduce time-varying damping factor related to image feature drift, through fuzzy logic algorithm, dynamically adjust the cooperative output of pressure-welding depth, offset the nonlinear amplification effect caused by amplitude attenuation, and combine the welding depth and interface fusion state fed back by real-time image, reconstruct the optimal parameter combination of pressure, welding depth and ultrasonic amplitude in the rolling optimization period, ensure the stability of heat input;

[0071] Step S4: Collect parameter drift data, post-welding interface image and quality detection feedback of each batch of welding, set iterative optimization method, based on image feature matching, analyze the correlation between image features and parameter drift, update process coupling model parameters using incremental learning, dynamically expand the identification range of critical region, for new materials or process fluctuation scenarios, compare the differences between new material surface image features and historical data, start online parameter identification process, real-time correct bifurcation boundary conditions, form a closed-loop iterative system including image monitoring, parameter control and model optimization, gradually enhance the adaptability to parameter critical point drift.

[0072] Specifically, the specific steps of the critical positioning method include:

[0073] Collect historical parameter data under dynamic working conditions, including basic data and dynamic images, and perform hierarchical processing on parameter data according to static basic values and dynamic fluctuations, such as decomposing the speed of the mechanical arm into rated speed and real-time fluctuation value, to facilitate subsequent analysis of the stability and fluctuation characteristics of the parameters, align the image sequences frame by frame to eliminate image displacement deviations caused by the movement of the mechanical arm, and then remove noise in the infrared image through Gaussian filtering to retain key temperature gradient features, thereby generating a hierarchical data set; wherein the basic parameters include but are not limited to the angle of rotation, the end movement speed / acceleration, the welding physical parameters, the vibration frequency, and the ultrasonic amplitude;

[0074] Perform cross-scale image feature parameter extraction. For visible light images of the material surface, perform super-resolution reconstruction to improve the image precision level for clear observation of the microstructure. Use the LBP algorithm to extract the surface texture entropy to reflect the uniformity of the surface roughness. Use the watershed algorithm to segment the surface micro-defects to calculate the density and distribution entropy of the defects to describe the aggregation degree of the defects. For infrared thermal imaging images, perform temperature field inversion based on the heat conduction equation to extract the highest temperature and temperature gradient slope of the interface. Align the infrared temperature field and the visible light image at the pixel level to establish a spatiotemporal mapping relationship containing the temperature gradient and the texture entropy, thereby realizing the correlation of different scale features, and clustering the feature parameters according to the material type to generate a feature parameter set containing the correlation weight between features;

[0075] At the physical mechanism layer, based on the ultrasonic vibration energy transmission equation and the composite material fusion dynamics model, the theoretical constraints between parameters are established. At the data-driven layer, cross-scale image feature parameters are introduced as soft constraints into the model. A physical information neural network architecture is used to build a process coupling model to predict the welding quality index. In the model training stage, historical welding data containing qualified and unqualified samples are used. The Bayesian optimization algorithm is used to adjust the weight coefficients of the image features to improve the response sensitivity of the process coupling model to image feature mutations, ensuring that the process coupling model can reflect the physical law and accurately capture the correlation between image features and welding quality. The process coupling model includes an input layer, a hidden layer, and an output layer. The input layer is used to receive each data set. The hidden layer embeds the physical mechanism constraint. The output layer is used to predict the welding quality index.

[0076] Set a basic critical threshold. For different batches of materials, collect 3 groups of verification samples for trial welding. Extract the image features and quality fluctuation data during the trial welding process. Calculate the sensitivity of the features and the quality of the batch materials through variance analysis. Dynamically correct the basic critical threshold based on the sensitivity, such as lowering the basic critical threshold when the sensitivity is high, to generate a critical positioning rule that contains a batched critical threshold and a three-dimensional judgment rule, such as a parameter combination less than the critical threshold for a batch of materials to be determined as far from the critical state.

[0077] Taking the robot arm speed, vibration frequency, and pressure as the three-dimensional coordinate axes, the Latin hypercube sampling method is used within the preset parameter range to generate multiple sets of parameter combinations, covering conventional and extreme working conditions to ensure the comprehensiveness of the sampling. Each set of parameter combinations is input into the process coupling model, and the corresponding welding quality index and image feature parameters are calculated. Combined with the critical threshold, the bifurcation point from stability to chaos is marked and annotated on the three-dimensional coordinate axis. The stable region, subcritical region, and chaotic region are marked with different colors. At the same time, the critical threshold surface of the image features is superimposed to form a parameter bifurcation map of the fusion parameters and image features. Among them, the parameter combinations in the stable region are all less than the stable critical threshold, and the parameter combinations in the subcritical region are between the stable critical threshold and the chaotic critical threshold. That is, the parameter combinations in the subcritical region are close to the chaotic critical threshold but have not entered the chaotic region. The welding quality begins to fluctuate slightly, and the image features show a transitional state that is a precursor to abnormalities. The parameter combinations in the chaotic region are greater than the chaotic critical threshold, indicating that the critical threshold has been broken through and the nonlinear chaotic region has entered. The welding quality fluctuates violently and there is a risk of serious defects.

[0078] For bifurcation points in the parameter bifurcation map, Monte Carlo simulation is used to generate multiple sets of perturbation parameter combinations within the preset parameter fluctuation neighborhood. The corresponding welding quality indicators are predicted through the process coupling model, and the critical threshold is combined to determine whether the current parameter combination is in a chaotic state. At the same time, multiple simulations are performed on the same parameter combination to calculate the probability of each parameter combination entering a chaotic state, obtain a baseline probability distribution, and then draw a probability cloud map on the three-dimensional coordinate axis;

[0079] For new materials, probability cloud maps of similar materials are retrieved from the historical database. A domain-adaptive transfer learning algorithm is used to transfer the probability distribution of similar materials to the new material, reducing the sampling volume of new materials and improving efficiency. At the same time, microscopic images of wear on the robotic arm joints are collected, and the correlation coefficient between the wear degree and the chaos probability is calculated and integrated into the probability cloud map to form a chaos boundary probability cloud map that simultaneously associates the material type and the equipment status. Among them, by extracting the characteristic parameter set of the new material and calculating the Euclidean distance with the characteristic vector of the material in the historical database, materials with a Euclidean distance less than the similarity difference threshold are defined as similar materials.

[0080] Based on the chaos boundary probability cloud map, three levels of chaotic areas are divided according to the chaos probability value, including low-risk area, warning area, and high-risk area. Each level of area corresponds to a different image feature threshold. The distribution law of image feature parameters in each area is statistically analyzed, and a correlation matrix containing parameter range, image features, and chaos probability is established. The test welding data is substituted into the correlation matrix, and the deviation between the actual chaos probability value and the predicted value is calculated. Once the deviation is greater than the correction threshold, the area boundary is fine-tuned to form a critical area parameter set to clarify the parameter range, corresponding image features, and chaos probability of different risk areas.

[0081] Specifically, the specific method of the critical early warning method comprises:

[0082] Start the high-resolution multispectral vision acquisition device to continuously acquire image data at a frequency synchronized with the welding rhythm, including images of the motion trajectory of the welding head at the end of the mechanical arm, real-time images of the welding interface, and images of the micro features of the material surface. Based on the inter-frame alignment algorithm, real-time correction of image displacement deviation caused by mechanical arm movement or vibration is performed by identifying fixed marker points in the images to ensure pixel-level alignment of images at different times. At the same time, real-time noise suppression is performed on the infrared images to retain temperature gradient features, generating a multispectral sequence;

[0083] Cross-scale image feature parameter extraction is performed on the multispectral sequence to generate a real-time feature sequence;

[0084] Based on the critical region parameter set, the stable critical threshold value corresponding to the current material batch is obtained, which is the boundary value between the stable region and the subcritical region. The real-time feature parameters are compared with the stable critical threshold value. If any feature exceeds the stable critical threshold value, it is marked as a potential abnormal feature. If all features are less than the stable critical threshold value, it is determined that the current welding is normal, and the next interval of real-time feature sequence is continuously acquired for further abnormality monitoring;

[0085] For potential abnormal features, the single-feature deviation rate is calculated based on the ratio of the difference between the real-time feature parameters and the stable critical threshold value to the stable critical threshold value. Combined with the feature correlation weight, weighted calculation is performed to obtain the comprehensive abnormality degree of the real-time feature sequence;

[0086] Combined with the process coupling model, the parameter combination of the real-time feature sequence is back calculated, and the region is located in the parameter bifurcation graph. The chaos probability and chaos probability change rate of the parameter combination are obtained by querying the chaos boundary probability cloud map;

[0087] If the comprehensive abnormality degree is greater than the abnormal threshold value and the chaos probability change rate is greater than the abnormal rate threshold value, it is determined that there is a critical drift, and a hierarchical early warning is performed for the critical drift. If the comprehensive abnormality degree is between the abnormal threshold value and the secondary abnormal threshold value and the parameter combination is located in the subcritical region, a first-level warning is triggered to prompt enhanced monitoring, and an abnormal feature time sequence curve is generated. Otherwise, a second-level warning is triggered to immediately start parameter reconstruction.

[0088] Specifically, the specific steps of the dynamic reconstruction method comprise:

[0089] Receive the second-level warning signal and the diagnostic data, perform multi-level analysis, extract the potential abnormal features and the abnormal feature time sequence curve contained in the warning, accurately locate the position of the current warning parameter combination in the parameter bifurcation graph based on the parameter bifurcation graph, and define it as a risk region, such as a high-risk region or a subcritical region;

[0090] Call the critical region parameter set, query the safety zone boundary parameter corresponding to the risk region, that is, the stable critical threshold, and through the process coupling model, the root cause of the abnormality is backstepped;

[0091] With the safety zone boundary as the target, the basic offset of each parameter in the parameter combination from the current value to the safety zone boundary is calculated. Due to the coupling relationship between parameters, such as the need to simultaneously increase the pressure to maintain energy input when the vibration frequency is reduced, the coupling coefficient is introduced to correct the basic offset, ensuring that the adjusted parameter combination still meets the energy balance. At the same time, combined with the material characteristics, the corresponding material correction coefficient is called to modify the offset again, avoiding the situation where the adjusted parameters are still at the critical edge. The smoothness of the mechanical arm motion trajectory is also monitored to avoid mechanical impact caused by parameter mutation;

[0092] The drift of the real-time characteristic parameter is calculated based on the difference between the real-time value and the stable critical threshold. The comprehensive drift is calculated by combining the load drift with the sensitivity as the weight. The basic coefficient is set for the comprehensive drift, and the drift acceleration is obtained synchronously. The time-varying damping factor is calculated by setting the dynamic coefficient;

[0093] The initial pressure and initial amplitude are obtained, and the time-varying damping factor is used for coordinated correction of the pressure and amplitude. The welding depth is monitored in real time through the visual device, and the target welding depth is called to calculate the target welding deviation. Once the target welding deviation is greater than the deviation threshold, the control parameters are adjusted based on the PID algorithm, and the compensation parameter set is finally generated;

[0094] Based on the image acquisition frequency, the rolling period is set. The dynamic objective function is constructed through the weighted calculation of temperature gradient fluctuation value, defect density growth rate, and fusion zone area proportion deviation. The weight is the feature correlation weight, and the heat input stability is the optimization objective. At the same time, the constraint conditions are set, such as the parameter adjustment amplitude in a single cycle not exceeding the safety range, to avoid violent fluctuations;

[0095] The particle swarm algorithm with mutation operator is used to search for the optimal parameter combination in the safety zone. The initial parameters are generated based on the compensation parameter set. Random mutation is introduced every iteration to avoid local optimum, thereby generating the optimal parameter combination for the current period. Linear interpolation is used for parameter smooth transition during adjustment to avoid new fluctuations caused by mutation;

[0096] The feature parameters of the three images after adjustment are collected, and the single feature deviation rate of each parameter in each image is calculated to obtain the average deviation rate of the stable critical threshold in the three images. If the average deviation rate is less than the deviation verification threshold, it is determined that the adjustment is initially effective, otherwise, it is determined that the adjustment is ineffective;

[0097] A plurality of rolling periods are continuously monitored, and the standard deviation of the parameter combination during the monitoring process is calculated. If the standard deviation is less than the standard verification threshold, it indicates that the trend is stable and there is no sharp fluctuation, and it is determined to be out of risk. Otherwise, it is determined to be not out of risk.

[0098] If any verification fails, that is, the adjustment is determined to be invalid or not out of risk, the above parameter adjustment and compensation process is repeated until all indicators meet the stable state requirements, thereby generating the optimal parameter combination and adjustment record.

[0099] Specifically, the specific steps of the iterative optimization method include:

[0100] Various key data during the welding process are automatically summarized and structured, and based on the welding batch, material type, and equipment state, three-level label classification is performed to ensure traceability. The parameter data is converted into dimensionless indicators, such as the percentage of the current value relative to the stable region boundary, and the quality data is converted into the deviation rate relative to the process standard value, and finally a full-dimensional database is formed. The key data includes the optimal parameter combination, the real-time drift curve during the parameter adjustment process, the abnormal feature time series data, the key image frame when the early warning is triggered, the detection data after welding, and the historical reference data.

[0101] The real-time image features during the welding process and the interface image features after welding are time-aligned, the dynamic time warping algorithm is used to calculate the similarity, the samples with similarity exceeding the correlation threshold are selected and defined as matching samples to ensure the consistency of process features and result features.

[0102] Based on the Pearson correlation coefficient, the linear correlation degree of the image feature drift and the parameter drift is calculated for the matching samples to analyze the linear relationship between the parameter drift and the image feature drift. Based on the mutual information entropy, the nonlinear correlation degree of the image feature drift and the parameter drift is calculated, and the influence weight of each parameter and image feature on the quality score is calculated through the random forest algorithm. Based on the correlation degree, the correlation rule library is sorted and generated, including the linear correlation coefficient, the nonlinear correlation coefficient, the quality influence weight, and the regulation and guidance rule.

[0103] Combined with the correlation rule library, the process coupling model is updated in layers. In the physical layer, the material coefficient in the ultrasonic vibration energy transmission equation is corrected according to the linear correlation rule between the parameter and the energy transmission. In the data-driven layer, the weight of the image feature in the neural network is adjusted based on the nonlinear correlation degree, and the three rules with the highest quality influence weight are converted into hard constraints of the model.

[0104] Incremental learning is carried out by using a federal learning framework, sub-models are trained in batches according to welding batches, a sub-model is trained for each batch of data, model parameters are reserved, sub-model parameters are fused by weighted average, historical data forgetting is avoided, and the prediction error of a new model on the last 3 batches of data needs to be less than the learning proportion of the original model, otherwise the weight is adjusted backtracking;

[0105] The historical data of different materials and different equipment states are selected as the verification set to ensure the prediction accuracy of the new model, and the optimization is confirmed to be effective after verification, so as to output the incremental optimized process coupling model;

[0106] The surface micro image and the thermodynamic parameter of the new material are collected, compared with the historical material feature library, a feature difference vector is constructed based on the difference between the real-time feature value and the historical mean value, and then divided by the historical standard deviation;

[0107] If the length of the feature difference vector is greater than the filing threshold, it is determined that the material is highly different, 3 groups of welding are carried out on the highly different material, the parameter drift and image feature data in the welding process are collected, the identified parameters such as thermodynamic parameters, fusion kinetics parameters and energy matching parameters are obtained by using the Bayesian filter to back-propagate the process coupling model, the identified parameters are substituted into the physical layer equation of the process coupling model, the temperature field prediction logic of the model is corrected, the key parameters of the model are identified in real time, the identified parameters and the feature difference vector are combined, the bifurcation boundary is corrected, the corresponding basic parameter component in the initial feature difference vector is replaced by the identified parameters, the length of the vector is recalculated, the equipment wear coefficient is obtained, and the boundary correction coefficient is calculated to correct the bifurcation boundary, and the corrected bifurcation boundary is adapted to the material properties and the equipment state, so that the dynamic adaptability of the boundary is ensured;

[0108] Otherwise, it is determined that the material is approximately matched, the bifurcation boundary and the parameters of the most similar material in the historical material feature library are reused as an initial scheme, 1 group of welding is carried out, the subtle differences between the welding and the historical data are compared, the bifurcation boundary is fine-tuned according to the feature difference vector and the associated weight, and the new material boundary is solidified after verification, otherwise the fine-tuning is repeated, and the efficiency and adaptability are balanced.

[0109] Embodiment 2:

[0110] Another embodiment provided by the present application is a composite material welding parameter control device, comprising a critical positioning module, a critical early warning module, a dynamic reconstruction module and an iterative optimization module.

[0111] The critical positioning module is used to establish a process coupling model containing pressure, welding depth, vibration frequency, ultrasonic amplitude, and image feature parameters based on the physical mechanism of mechanical arm motion characteristics, ultrasonic vibration transmission path, and material response, combined with image features of the welding interface and material surface; through acquisition of historical parameter data and hierarchical processing, extraction of cross-scale image features, and setting of critical threshold, a parameter bifurcation atlas is generated, and a chaotic boundary probability cloud map is constructed in combination with material image feature differences to locate the parameter critical region.

[0112] The critical early warning module is used to continuously acquire image data of the motion trajectory of the welding head at the end of the mechanical arm, the state of the welding interface, and the micro features of the material surface through a high-resolution multispectral vision acquisition device, and to extract real-time features after correction and noise reduction processing; based on the parameter set of the critical region, abnormal changes in the features are analyzed, and nonlinear precursors of parameters close to critical values are captured to trigger graded early warnings.

[0113] The dynamic reconstruction module is used to calculate the safety offset of the current parameters to the subcritical region based on the parameter bifurcation atlas after triggering the early warning, and to correct the offset in combination with the parameter coupling relationship; a time-varying damping factor related to the image feature drift amount is introduced, and the pressure-welding depth collaborative output is dynamically adjusted through a fuzzy logic algorithm to offset the nonlinear amplification effect; the optimal parameter combination is reconstructed in a rolling optimization period to ensure the stability of heat input.

[0114] The iterative optimization module is used to collect parameter drift, interface image, and quality detection data of each batch of welding, and to mine the association rules of image features and parameter drift after structured processing; the process coupling model is updated using incremental learning, and the bifurcation boundary is corrected in real time by comparing feature differences and online parameter identification for new materials or process fluctuations to form a closed-loop system of image monitoring, parameter control, and model optimization, thereby enhancing the adaptability to parameter critical point drift.

[0115] Working principle and effects:

[0116] Based on the motion characteristics of the mechanical arm, the transmission law of ultrasonic vibration and the response mechanism of materials, the multi-parameter process coupling model is established by fusing the cross-scale image features of the welding interface and surface. The parameter bifurcation atlas and chaos boundary probability cloud map are generated through simulation to clearly define the parameter boundaries of stable region, subcritical region and chaotic region. The high-resolution visual device is used to collect image data in real time, extract key features and analyze abnormal changes, capture the nonlinear precursor of parameters approaching critical values, and trigger hierarchical warning. After the warning, the safety offset is calculated and the time-varying damping factor is introduced to dynamically adjust the parameters such as the speed of the mechanical arm, the amplitude of the ultrasonic vibration and the pressure, and to migrate the parameter combination to the safe region to offset the nonlinear amplification effect. At the same time, the whole process data is collected to mine the correlation law of image features and parameter drift, and the model is updated through incremental learning to dynamically correct the critical boundary for new materials or process fluctuations, forming an adaptive closed loop. The critical point drift problem that the traditional method cannot handle is effectively solved, and through accurate identification of chaos precursors, dynamic regulation of parameters and continuous optimization of the model, the dramatic fluctuation of heat input caused by the parameter breaking through the critical value is avoided, and the defects such as overheating carbonization and incomplete fusion are inhibited, which significantly improves the stability of the welding quality and ensures that the strength of the welding interface meets the standard, effectively avoiding the risk of mechanical performance degradation caused by chaos effect.

[0117] The above only describes the preferred embodiments of the present application, and the protection scope of the present application is not limited to the above-mentioned embodiments. Any technical solutions falling within the concept of the present application shall fall within the protection scope of the present application. It should be noted that for ordinary technical personnel in the technical field, some improvements and decorations without departing from the principles of the present application shall also be considered as the protection scope of the present application.

Claims

1. A method for controlling welding parameters of composite materials, characterized in that: include: Build a process coupling model, set a critical location method, generate a parameter bifurcation map, and construct a chaotic boundary probability cloud map to correlate the image feature differences of different materials and locate the critical parameter area; Acquire multispectral sequences, set critical warning methods, extract key features, analyze abnormal changes in image features, and trigger pre-alarms once abnormal changes are identified; Calculate the safety offset, set the dynamic reconstruction method, introduce the time-varying damping factor, and dynamically adjust the collaborative output. Combined with the real-time image feedback of the welding depth and interface fusion state, reconstruct the optimal parameter combination; Collect batch welding data, set up iterative optimization methods, update process coupling model parameters, dynamically expand the critical area identification range, and correct bifurcation boundary conditions in real time.

2. A composite material welding parameter control method according to claim 1, characterized in that: The critical positioning method includes: Collect historical parameter data under dynamic working conditions, perform hierarchical processing, and perform inter-frame alignment and Gaussian filtering on the image sequence to generate a hierarchical data set; Extract cross-scale image feature parameters, perform super-resolution reconstruction of visible light images of material surfaces, extract surface texture entropy, segment surface micro-defects and calculate the density and distribution entropy of defects, perform temperature field inversion on infrared thermal imaging images based on the heat conduction equation, extract the maximum temperature and temperature gradient slope of the interface, cluster feature parameters by material type, and generate a feature parameter set; Based on the ultrasonic vibration energy transmission equation and the composite material fusion dynamics model, theoretical constraints between parameters are established. Cross-scale image feature parameters are introduced into the model as soft constraints to construct a process coupling model. The weight coefficients of image features are adjusted using the Bayesian optimization algorithm using historical welding data to train the process coupling model. The basic critical threshold is set, the image features and quality fluctuation data during the test welding process are extracted, the sensitivity is calculated, the basic critical threshold is dynamically corrected, and the critical positioning rules are generated.

3. A composite material welding parameter control method according to claim 2, characterized in that: The critical positioning method further includes: Construct a three-dimensional coordinate axis, generate multiple sets of parameter combinations within the preset parameter range, combine with the process coupling model, mark the bifurcation point, and divide it into stable region, subcritical region, and chaotic region to generate a parameter bifurcation map; For bifurcation points in the parameter bifurcation map, multiple sets of disturbance parameters are generated, a baseline probability distribution is calculated using the process coupling model, and a probability cloud diagram is drawn on a three-dimensional coordinate axis; For new materials, probability cloud maps of similar materials are retrieved from the historical database, and the probability distribution of similar materials is transferred to the new material using a domain-adaptive transfer learning algorithm. The new material's characteristic parameter set is extracted and the Euclidean distance is calculated with the characteristic vectors of the materials in the historical database. Materials with a Euclidean distance less than a similarity difference threshold are defined as similar materials. Collect microscopic images of the joint wear of the robotic arm and calculate the correlation coefficient, which is then integrated into the probability cloud map to form a chaotic boundary probability cloud map; Based on the chaos boundary probability cloud map, the chaos is divided into low-risk areas, warning areas, and high-risk areas according to the chaos probability value, and the distribution law of image feature parameters in each area is counted to establish a correlation matrix; The deviation between the actual chaotic probability value and the predicted value is calculated. When the deviation is greater than the correction threshold, the region boundary is adjusted to form a critical region parameter set.

4. A composite material welding parameter control method according to claim 3, characterized in that: The critical warning method includes: Collect image data in real time, extract cross-scale image feature parameters, and generate real-time feature sequences; Based on the critical region parameter set, a stable critical threshold is obtained. If any feature in the real-time feature sequence exceeds the stable critical threshold, it is marked as a potential abnormal feature; otherwise, it is determined to be normal and continuous abnormality monitoring is performed; For the potential abnormal features, the single feature deviation rate is calculated and weighted calculation is performed to obtain the comprehensive abnormality degree of the real-time feature sequence.

5. A composite material welding parameter control method according to claim 4, characterized in that: The critical warning method further includes: In combination with the process coupling model, the parameter combination is inferred, and the region where the parameter is located is located in the parameter bifurcation map. At the same time, the chaos boundary probability cloud map is queried to obtain the chaos probability change rate; If the comprehensive abnormality is greater than the abnormality threshold and the chaotic probability change rate is greater than the abnormality rate threshold, it is determined to be a critical drift; A graded warning is performed on critical drift. If the comprehensive abnormality is between the abnormal threshold and the secondary abnormal threshold and the parameter combination is in the subcritical area, a first-level warning is triggered and an abnormal characteristic time series curve is generated. Otherwise, a second-level warning is triggered.

6. A composite material welding parameter control method according to claim 5, characterized in that: The dynamic reconstruction method comprises: receiving a secondary warning signal, locating a position of the warning parameter combination in the parameter bifurcation map, defining it as a risk area, obtaining a stability critical threshold of the risk area, and calculating a basic offset; In combination with the correlation matrix, a coupling coefficient is introduced to correct the basic offset, and at the same time, a material correction coefficient is called to correct the basic offset again; Calculate the drift of real-time characteristic parameters, use sensitivity as weight, and combine it with load drift to calculate the comprehensive drift; Obtain the drift acceleration, combine it with the basic coefficient and dynamic coefficient, calculate the time-varying damping factor, and coordinately correct the initial pressure and initial amplitude; The real-time welding depth is obtained, the target welding depth is called, and the target welding deviation is calculated. Once the target welding deviation is greater than the deviation threshold, the control parameters are adjusted based on the PID algorithm to generate a compensation parameter set.

7. A composite material welding parameter control method according to claim 6, characterized in that: The dynamic reconstruction method further includes: Set the rolling cycle, build a dynamic objective function, and set constraints; A particle swarm algorithm with a mutation operator is used to search for the optimal parameter combination within the safe zone and adjust the parameters; Collect the characteristic parameters of the three frames of images after adjustment and calculate the average deviation rate; If the average deviation rate is less than the deviation verification threshold, the adjustment is considered valid; otherwise, the adjustment is considered invalid. Continuously monitor multiple rolling periods and calculate the standard deviation. If the standard deviation is less than the standard verification threshold, it is determined to be out of risk; otherwise, it is determined to be in risk. If any verification fails, the parameter adjustment and compensation process is repeated until all indicators meet the steady-state requirements and the optimal parameter combination is generated.

8. A composite material welding parameter control method according to claim 7, characterized in that: The iterative optimization method comprises: Summarize various key data in the welding process and perform structured processing on the key data; Perform temporal alignment on the real-time image features during welding and the image features of the interface after welding, calculate the similarity, and filter out matching samples whose similarity exceeds the correlation threshold; For matching samples, the linear correlation and nonlinear correlation between image feature drift and parameter drift are calculated, and the quality impact weight is calculated to generate an association rule base. The process coupling model is updated hierarchically, and incremental learning is performed using a federated learning framework. The sub-models are trained in blocks according to welding batches, and the sub-model parameters are fused through weighted averaging. A validation set is constructed to output the incrementally optimized process coupling model.

9. A composite material welding parameter control method according to claim 8, characterized in that: The iterative optimization method further comprises: Collect relevant parameters of new materials, compare them with historical material feature libraries, and construct feature difference vectors; If the modulus of the characteristic difference vector is greater than the archiving threshold, it is determined to be a high-difference material, a test welding group is established, test welding parameters are collected, and identification parameters are reversed. The process coupling model and the characteristic difference vector are updated based on the identification parameters to correct the bifurcation boundary; Otherwise, it is determined to be an approximately matching material, the initial solution is called, and the bifurcation boundary is fine-tuned.

10. A composite material welding parameter control device, used to implement a composite material welding parameter control method according to any one of claims 1 to 9, characterized in that: include: Critical positioning module, critical warning module, dynamic reconstruction module and iterative optimization module; The critical positioning module is used to build a process coupling model, generate a parameter bifurcation map, and construct a chaotic boundary probability cloud map to associate the image feature differences of different materials and locate the critical area of ​​parameters; The critical warning module is used to obtain multi-spectral sequences, extract key features, analyze abnormal changes in image features, and trigger a pre-alarm once an abnormal change is identified; The dynamic reconstruction module is used to calculate the safety offset, dynamically adjust the collaborative output, and reconstruct the optimal parameter combination based on the welding depth and interface fusion state fed back by real-time images; The iterative optimization module is used to collect batch welding data, update process coupling model parameters, dynamically expand the critical area identification range, and correct bifurcation boundary conditions in real time.

Images