Multi-dimensional self-adaptive conductive connecting piece welding positioning system

Abstract

The invention discloses a multi-dimensional self-adaptive conductive connecting piece welding positioning system, and relates to the technical field of conductive connecting piece intelligent welding positioning, the system comprises a multi-dimensional data acquisition module, and a multi-modal sensor is used for acquiring contour, feature, current and other data; the data processing and analysis module processes data through deep learning and a physical model and predicts deformation; the self-adaptive positioning adjusting module controls the eight-axis linkage mechanical arm to compensate deformation through MPC; the welding process monitoring module fuses multi-source data monitoring and triggers adjustment when abnormity occurs; and the intelligent decision and feedback module realizes closed-loop control and optimization through knowledge graph and active learning decision feedback. Through multi-modal sensing and an intelligent algorithm, accurate positioning and dynamic compensation of welding deformation are achieved, welding parameters are adjusted in a self-adaptive mode, intelligent decision making and remote operation and maintenance are combined, the welding quality and stability are improved, errors and cost are reduced, and powerful support is provided for efficient welding production.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent welding positioning of conductive connectors, and particularly to a multi-dimensional adaptive welding positioning system for conductive connectors. Background Art

[0002] In the fields of modern electronic devices, power systems, etc., the welding quality of conductive connectors plays a crucial role in the overall performance and safety. Traditional welding of conductive connectors mostly relies on manual operation. During the welding process, workers need to manually adjust the position of the connectors and welding parameters based on experience. This method is not only inefficient and difficult to meet the requirements of large-scale production, but also greatly affected by human factors. The operating levels of different workers vary, resulting in unstable welding quality and prone to problems such as false soldering and short circuits, affecting the reliability and service life of products.

[0003] With the development of automation technology, some enterprises have adopted automated welding equipment for welding conductive connectors. However, existing automated welding positioning systems have many limitations. On the one hand, their positioning accuracy is difficult to meet the requirements of high-precision welding. Traditional positioning systems mostly use a single type of sensor, such as only using a vision sensor or a displacement sensor, and cannot comprehensively sense the position, shape, and attitude information of conductive connectors. When facing connectors with complex shapes or reflective surfaces, the positioning error is relatively large. Moreover, these systems lack an effective compensation mechanism for factors such as thermal deformation and electromagnetic interference during the welding process. During the welding process, the tiny deformation of the connectors due to heat, and the interference of the complex electromagnetic environment in the workshop on the sensors, will both cause a deviation between the actual welding position and the ideal position, reducing the welding quality.

[0004] On the other hand, the parameter adjustment and decision-making capabilities of existing welding positioning systems are insufficient. The setting of welding parameters (such as welding current, time, etc.) is usually fixed, or only simply classified and set according to the type of connectors, and cannot be adjusted in real time and dynamically according to the actual welding situation. When the material, size of the connectors change, or the environmental temperature, humidity, and electromagnetic environment fluctuate, the fixed welding parameters are difficult to ensure the welding quality. At the same time, the system lacks intelligent decision-making capabilities. When welding deviations or failures occur, it cannot quickly and accurately analyze the reasons and take effective solutions, often requiring manual intervention for troubleshooting, increasing the production downtime and maintenance costs, and being difficult to meet the development needs of modern intelligent manufacturing. Summary of the Invention

[0005] The multi-dimensional adaptive welding positioning system for conductive connectors proposed by the present invention is to solve the problems mentioned in the above existing technologies.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] A multi-dimensional adaptive conductive connector welding positioning system includes the following modules:

[0008] Multi-dimensional data acquisition module: At least 5 laser displacement sensors are deployed on the welding workbench, and through the triangulation optimization algorithm , where d is the measurement distance and L is the sensor baseline length, is the fixed emission angle, is the reflected light reception angle. A Rogowski coil current sensor is set in the welding circuit to monitor current changes. The temperature and humidity sensor uses an integrated sensor. All sensor data is transmitted to the data processing module;

[0009] Data processing and analysis module: Use the multi-scale feature fusion network MSFF-Net based on deep learning to process data. For laser displacement data, use the adaptive wavelet threshold denoising algorithm, and the formula is , is the threshold of the i-th scale, is the noise standard deviation, is the number of data points at this scale;

[0010] Adaptive positioning adjustment module: The multi-dimensional motion platform uses an eight-axis linkage robotic arm. The MPC algorithm plans the motion trajectory of the robotic arm in advance, and optimizes the control input through the formula , where u is the control input sequence, is the predicted output, is the reference output, Q and R are weight matrices. The robotic arm adopts a flexible joint design and adaptively adjusts the joint stiffness according to the shape and force condition of the connector;

[0011] Welding process monitoring module: Comprehensively monitor the welding process with multi-source data, and use the state estimation method based on Kalman filtering. The formula is , is the updated state estimation, is the predicted state estimation, is the Kalman gain, is the measured value, H is the measurement matrix. At the same time, use the deep reinforcement learning algorithm to evaluate and optimize the welding process in real time;

[0012] Intelligent decision-making and feedback module: Use the intelligent decision-making system based on the knowledge graph, conduct knowledge reasoning and mining through the graph neural network. When the system has deviations or failures, the intelligent decision-making system analyzes the reasons and generates processing strategies. At the same time, the system adopts an active learning mechanism to update the knowledge graph and transmits the decision-making results in a timely manner through a real-time feedback mechanism.

[0013] Furthermore, it also includes an electromagnetic environment perception and compensation module, which arranges electromagnetic sensors around the system to monitor the electromagnetic environment intensity, frequency and direction in real time. By establishing an electromagnetic interference model, it analyzes the impact of electromagnetic interference on sensor data acquisition and system control, and uses the formula Perform electromagnetic interference compensation, where is the electric field strength after compensation, is the original electric field strength, is the compensation coefficient, is the electric field strength of each interference source, is the interference source frequency, At the same time, the system adopts an adaptive electromagnetic shielding strategy to dynamically adjust the parameters and structure of the shielding layer according to changes in the electromagnetic environment.

[0014] Furthermore, it also includes a welding parameter dynamic optimization module, establishes a welding parameter optimization model based on a fuzzy neural network, fuzzifies the input parameters through a fuzzy inference system, automatically adjusts the fuzzy rules and membership functions using the learning ability of the neural network, and optimizes the model using a genetic simulated annealing algorithm. Controlled annealing process, where is the kth generation temperature, For the temperature reduction coefficient, during the welding process, the welding parameters are dynamically adjusted according to the real-time monitoring data and changes in the welding status.

[0015] Furthermore, the laser displacement sensor in the multi-dimensional data acquisition module adopts dual-wavelength interferometry measurement technology, which measures with two lasers of different wavelengths, and uses the formula Detects small shape changes and defects on the surface of the connection parts, including To measure distance changes, and For two laser wavelengths, is the number of interference fringe changes. At the same time, the sensor adopts an adaptive measurement range adjustment strategy to automatically adjust the measurement range according to the approximate shape and size of the connector.

[0016] Furthermore, in the data processing and analysis module, for visual images, a dynamic physical model of the conductive connector is established and the formula Evaluate the influence of welding heat on positioning, where Q is the total heat effect, m is the mass of the connection, and c is the specific heat capacity. is the temperature change, I is the welding current, R is the resistance, t is the welding time, is the influence coefficient of each environmental factor, It is the quantization value corresponding to environmental factors. This module processes visual images using an image enhancement technology based on a generative adversarial network. The generator network generates high-quality enhanced images based on the input low-quality images, and the discriminator network judges the authenticity of the generated images. At the same time, a feature extraction algorithm based on geometric feature matching is adopted, combined with the geometric invariant features of the images, to locate the key feature points of the connecting piece.

[0017] Furthermore, the multi-dimensional motion platform of the adaptive positioning and adjustment module adopts magnetic levitation support technology to reduce mechanical friction and vibration. At the same time, the motion platform is equipped with a force sensor to monitor the force between the connecting piece and the welding tool in real time, and adjusts the motion and posture of the robotic arm through a force feedback control algorithm to ensure uniform force on the connecting piece during the welding process.

[0018] Furthermore, the welding process monitoring module adopts a temperature monitoring system based on fiber optic sensing technology. Fiber optic sensors are distributed around the welding area to monitor the temperature changes during the welding process. By establishing a temperature field model, the influence of temperature distribution on welding quality and connecting piece deformation is analyzed, and the formula is used to predict the temperature change trend, where T is the temperature, q is the heat source intensity, k is the thermal conductivity, is the thermal diffusivity.

[0019] Furthermore, the electromagnetic environment perception and compensation module adopts a distributed electromagnetic sensor network to achieve data sharing and collaborative work among sensors through wireless communication technology. At the same time, the system adopts an electromagnetic interference prediction algorithm based on machine learning to predict the occurrence probability and intensity of electromagnetic interference in advance according to historical electromagnetic environment data and current meteorological conditions and other factors.

[0020] Furthermore, the welding parameter dynamic optimization module adopts a multi-objective optimization strategy. By introducing the Pareto optimal theory, a set of non-dominated solutions is generated, and the user selects the welding parameter combination according to actual needs. At the same time, the system adopts an online learning mechanism to adjust the optimization model parameters according to the real-time welding results.

[0021] Furthermore, the system also includes a remote collaboration and intelligent diagnosis module to achieve remote collaboration and data sharing among multiple welding positioning systems through an industrial Internet platform. Welding data, process parameters, and fault information are exchanged in real time among different systems. At the same time, the system adopts an intelligent diagnosis algorithm based on deep learning to monitor the operation status of the system and diagnose faults in real time. By analyzing sensor data and system logs, the fault cause is located, and corresponding repair suggestions and solutions are provided. The operator can view the system status, receive fault alarms, and perform remote operations in real time through a mobile terminal or a remote monitoring center.

[0022] Compared with the existing technologies, the beneficial effects of the present invention are:

[0023] In terms of positioning accuracy, through the distributed layout of multi-modal sensors and unique measurement algorithms, it is able to comprehensively and accurately collect information such as the three-dimensional contour and surface features of conductive connectors. Combined with the comprehensive compensation for factors such as thermal deformation and electromagnetic interference, the positioning error is effectively reduced, ensuring the accuracy of the welding position and improving the consistency of welding quality.

[0024] In terms of welding process control, the system can dynamically optimize welding parameters using intelligent algorithms based on the real-time monitored connector status, environmental parameters, and welding quality data, achieving adaptive adjustment of the welding process. Whether it is the change in the material and size of the connector or the fluctuation of environmental conditions, the system can quickly adjust the welding parameters to ensure stable and reliable welding quality, avoid problems such as false soldering and welding through caused by improper parameters, and improve the qualified rate of products.

[0025] The intelligent decision-making and feedback mechanism of this system also has great advantages. Through the knowledge graph and deep learning algorithms, the system can quickly analyze the deviations and fault causes that occur during the welding process and generate effective solutions in a timely manner to achieve closed-loop control. At the same time, the remote collaboration and intelligent diagnosis functions enable operators to grasp the system status in real time, perform remote operation and maintenance, greatly reducing production downtime and lowering operation and maintenance costs. In addition, the system also has the ability of online learning and self-optimization. With the increase in usage time and the accumulation of data, its performance and decision-making accuracy will continue to improve, providing strong technical support for enterprises to achieve high-efficiency and high-quality welding production of conductive connectors. Brief Description of the Drawings

[0026] Figure 1 It is a schematic block diagram of the multi-dimensional adaptive conductive connector welding positioning system proposed by the present invention;

[0027] Figure 2 It is a column chart comparing the positioning errors of the traditional system, the improved system, and this system under different welding batches;

[0028] Figure 3 It is a line chart showing the change of the welding qualified rate over time during the continuous 5-day production process of the traditional system and this system;

[0029] Figure 4 It is a bar chart comparing the production efficiencies of the traditional system, the improved system, and this system. Detailed Embodiment

[0030] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like to indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0032] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the said features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below with reference to the accompanying drawings.

[0033] Reference Figures 1 to 4 : A multi-dimensional adaptive conductive connector welding positioning system, including the following modules:

[0034] Multi-dimensional data acquisition module: Four Pepperl+Fuchs ODSL-1000 laser displacement sensors are installed vertically at the four corners of the welding workbench, and one sensor of the same model is installed at the center at a 45° angle. All sensors are fixed with aluminum alloy brackets, and shock-absorbing rubber pads are installed at the bottom of the brackets to reduce external vibration interference. The measurement beams of the sensors overlap each other to form a three-dimensional measurement area. At the initial startup of the system, the laser displacement sensor is calibrated using a high-precision calibration block (dimensional accuracy ±0.001mm) to determine the baseline length L. During operation, the sensor collects data at a frequency of 250Hz and uses the triangulation optimization algorithm to calculate the optimal measurement range. Calculate the distance from the surface of the conductive connection component, where the fixed emission angle is 45°, and the reflected light reception angle is measured in real time by a high-precision angle encoder (resolution 0.001°) built into the sensor, ensuring that the measurement accuracy reaches ±0.008 mm.

[0035] The vision camera selects the DAheng Image MER-162-30UM-LX model with a polarization filter and is installed on a dedicated adjustable cloud platform directly above the workbench. By adjusting the cloud platform, the horizontal, vertical movement and angle rotation of the camera can be realized. During the installation and commissioning phase, the camera is calibrated for internal and external parameters using a checkerboard calibration board to ensure the accuracy of image acquisition. During operation, the camera acquires images at a frame rate of 35 fps and a resolution of 16 million pixels. The polarization filter is driven by a stepper motor and can be adjusted within the range of 0 - 180°. According to the ambient light intensity on site and the reflection of the connection component surface, the filter angle is automatically adjusted to effectively eliminate the reflection interference.

[0036] The Rogowski coil current sensor is connected in series in the welding circuit. To avoid electromagnetic interference, the sensor signal line uses a double-layer shielded wire and is filtered using a magnetic ring. The weak current signal output by the sensor is converted into a standard voltage signal of 0 - 5V by the signal conditioning circuit and then converted into a digital signal by a high-precision ADC (16-bit resolution) and input into the data acquisition card. The Vaisala HMT330 temperature and humidity sensor is installed on a column 1 meter away from the welding workbench. The probe part of the sensor is protected by a protective cover to prevent damage from welding spatter. The ambient temperature and humidity data are collected at a frequency of once per minute, and the accuracies reach ±0.3°C and ±1.5%RH respectively. The data collected by all sensors are transmitted to the data processing module through the Profibus-DP fieldbus according to the master-slave communication protocol at a transmission rate of 12 Mbps. A repeater is installed every 10 meters on the bus to ensure the stability and reliability of data transmission.

[0037] Data processing and analysis module: The data processing module uses a high-performance industrial server equipped with an Intel Xeon W-3275M processor, 128 GB of memory, and an NVIDIA A100 graphics card, running the Ubuntu 20.04 operating system. After the laser displacement data enters the system, it first undergoes adaptive wavelet threshold denoising processing. The median method is used to calculate the noise standard deviation , where is the original data sequence, is the median of the data sequence. According to the characteristics of different data segments, according to the formula (N iDynamically adjust the threshold according to the number of points in this data segment to denoise the data. The denoised data is then processed by smooth filtering, using the five-point cubic smoothing algorithm to further improve the smoothness of the data.

[0038] Visual images are processed through a multi-scale feature fusion network (MSFF-Net) based on deep learning. The network contains 5 convolutional layers of different scales. The first convolutional layer uses a 3×3 convolutional kernel with a stride of 1 to extract the basic features of the image; the subsequent convolutional layers gradually increase the convolutional kernel size to 7×7, and the stride is adjusted accordingly to obtain features of different scales. Atrous convolutional layers are added after each convolutional layer to expand the receptive field while reducing the number of parameters. Spatial attention mechanism and channel attention mechanism modules are introduced. The spatial attention mechanism calculates the weights at different positions of the image to highlight key regions; the channel attention mechanism weights different feature channels to enhance the response of important feature channels. The network is trained with a large amount of labeled conductive connector image data, using the Adam optimizer, with the learning rate set to 0.001 and trained for 100 epochs to improve the recognition ability of connector features.

[0039] Evaluate the welding heat influence by establishing a dynamic physical model of the conductive connector. For connectors of different materials (such as copper and aluminum alloy), their specific heat capacity c, density and other parameters are retrieved from the built-in material database, which contains the thermophysical property parameters of common conductive materials. The resistance R of the connector is measured by the four-probe method before welding, and the measuring device uses a Keithley 2400 source meter. The environmental factor influence coefficient k i is determined by performing multiple linear regression analysis on historical welding data, where factors such as environmental temperature and humidity, and electromagnetic interference are used as independent variables, and the degree of welding heat influence is used as the dependent variable. Use the formula to calculate the total heat influence and predict the welding deformation trend. At the same time, combine a finite element analysis software (such as ANSYS) to verify and correct the prediction results to improve the accuracy of the prediction.

[0040] Adaptive positioning adjustment module: The multi-dimensional motion platform uses a KUKA KR210R2700extra eight-axis linkage robotic arm. A customized pneumatic fixture is installed at the end of the robotic arm, and the surface of the fixture is made of insulating rubber material to prevent damage to the conductive connector during clamping. Renishaw RGH22Z type grating scales (resolution 0.1μm) and absolute encoders (24-bit resolution) are installed at each joint of the robotic arm to real-time feedback the position and angle information of the joints. During the installation and commissioning stage of the robotic arm, a laser interferometer (such as Renishaw XL-80) is used to calibrate its positioning accuracy, and by adjusting the joint parameters and compensation algorithms, ensure that the positioning accuracy reaches ±0.003mm.

[0041] The manipulator is controlled based on the Model Predictive Control (MPC) algorithm. At the beginning of each control cycle (set to 10 ms), the MPC algorithm constructs an optimization problem according to the predicted welding deformation (provided by the data processing and analysis module), the current position and attitude of the manipulator (obtained through grating rulers and encoders), and the preset target position. The objective function is , where the prediction horizon N is set to 15 and the control horizon is set to 5. The weight matrices Q and R are optimized and adjusted through simulation tests and actual debugging to balance the tracking error and control energy. When solving the optimization problem, the interior point method is used for solution, and the optimal control input sequence u is obtained through iterative calculation to drive the joint motors of the manipulator (the servo motors use the Siemens 1FT6 series) to move.

[0042] The joints of the manipulator adopt a flexible joint design, which is internally composed of a high-precision harmonic reducer, elastic elements and dampers. During the process of gripping and moving the connecting piece, the force sensors at the joints (using the ATINano17 series with a resolution of 0.01 N) monitor the force in real time. When the force exceeds the preset threshold, the control system automatically changes the joint stiffness by adjusting the pre-tightening force of the elastic element and the damping coefficient of the damper according to the magnitude and direction of the force, so as to adapt to connecting pieces of different shapes and weights and avoid deformation or damage of the connecting piece caused by excessive rigidity.

[0043] Welding process monitoring module: Eight fiber optic temperature sensors (using the TF-800 series of Lumasense Technologies) are evenly arranged around the welding area in the circumferential direction. The probes of the sensors are pasted on the surface of the connecting piece 2 mm away from the welding point through high-temperature resistant glue, and a ceramic protection sleeve is sleeved outside the probes to prevent damage by welding spatter. The temperature measurement principle of the fiber optic temperature sensor is based on the Raman scattering effect of the optical fiber, which can monitor the temperature change during the welding process in real time, with an accuracy of up to ±0.1 °C and a sampling frequency of 10 Hz.

[0044] Combined with the data collected by the laser displacement sensor, vision camera and current sensor, a multi-sensor fusion monitoring architecture is constructed. The state estimation method based on the Kalman filter is used to estimate the position and attitude of the connecting piece in real time. In the system initialization stage, the state vector contains the position of the connecting piece , attitude (rotation angles around the x, y, and z axes ), as well as velocity , angular velocity information, and the covariance matrix is initialized and assigned according to the measurement accuracy of the sensors. At each time step k, according to the formula Update the state estimation, where the measurement matrix H is determined according to the measurement parameters and installation positions of different sensors. For example, the laser displacement sensor measures the distance, corresponding to the position information in the state vector; the vision camera identifies the feature points, corresponding to the position and attitude information.

[0045] When it is monitored that the deviation between the actual position and the preset position exceeds ±0.08 mm, or the current fluctuation exceeds ±5% of the rated value, or the temperature changes abnormally (such as the heating rate exceeds the preset threshold), the adaptive adjustment mechanism is immediately triggered. Through the deep reinforcement learning algorithm, taking the state during the welding process (including information such as position deviation, current, and temperature) as the input, the optimal adjustment action is selected from the preset action set (such as adjusting the position of the robotic arm, changing the welding current, pausing the welding, etc.). The deep reinforcement learning model adopts the deep Q-network (DQN) architecture, and through interaction with the environment (the actual welding process), continuously learns and optimizes the strategy to achieve the best welding process control.

[0046] Intelligent decision-making and feedback module: The intelligent decision-making system is constructed based on a knowledge graph. The construction process of the knowledge graph is as follows: First, knowledge is extracted from a large number of welding process manuals, material standards, academic papers, and historical fault cases within the enterprise through natural language processing technology, and this knowledge is transformed into a structured triple (entity-relationship-entity) form. Then, a graph database (such as Neo4j) is used to store and manage this knowledge to construct the knowledge graph. Through the graph neural network (GNN), reasoning and mining are performed on the knowledge graph. When there are deviations or faults in the system, the GNN algorithm searches for relevant knowledge nodes and relationship paths in the knowledge graph according to the current abnormal state information to quickly analyze the reasons. The system adopts an active learning mechanism. When new welding data is generated, the data is first cleaned to remove outliers and noisy data. Then, key features are extracted through a feature extraction algorithm (such as principal component analysis PCA), and the processed data is added to the training dataset. The knowledge graph is updated and optimized regularly (once a week). According to the knowledge information in the new data, knowledge nodes and relationships are added or modified to improve the accuracy of decision-making. The decision result is sent to the adaptive positioning adjustment module and the welding parameter optimization module through a real-time feedback mechanism. The feedback communication uses the Socket communication method based on the TCP / IP protocol to ensure the real-time and reliability of data transmission. During the feedback process, the decision-making instructions are encrypted (using the AES-128 encryption algorithm) to prevent the instructions from being tampered with or stolen during transmission, realizing the closed-loop control and continuous optimization of the system.

[0047] In the present invention, there is also an electromagnetic environment perception and compensation module: At the four corners of the ceiling, the middle of the walls, and the corners of the floor in the welding workshop, 8 P6015A high-voltage probes attached to Tektronix TDS3054C oscilloscopes are evenly arranged as electromagnetic sensors. These probes send the monitored electromagnetic environment data to the central processing unit through a wireless communication module (using the ZigBee protocol, operating frequency band 2.4 GHz). The central processing unit establishes an electromagnetic interference model, analyzes the historical electromagnetic interference data and the sensor measurement data, and uses the formula to calculate the compensated electric field strength. Among them, the compensation coefficient is obtained by fitting the historical data through the least squares method, , , are respectively the electric field strength, frequency, and phase of each interference source, which are obtained by analyzing the electromagnetic signals through a spectrum analyzer (such as Keysight N9020A).

[0048] The system adopts an adaptive electromagnetic shielding strategy, and a liftable electromagnetic shielding curtain is set around the welding workbench. The shielding curtain is composed of two layers of metal meshes (copper mesh, mesh number 200) with an insulating material (polyvinyl chloride) sandwiched in the middle, and the lifting of the shielding curtain is realized by a servo motor driving a chain. When the electromagnetic sensor detects that the electromagnetic interference intensity exceeds the preset threshold, the control system adjusts the height and angle of the shielding curtain according to the frequency and direction of the interference to achieve the best shielding effect. At the same time, the key electronic components of the system (such as sensors, data acquisition cards, etc.) are individually shielded by metal shielding boxes, and the shielding boxes are connected to the grounding system of the workshop through grounding wires to further reduce the impact of electromagnetic interference on the system.

[0049] In the present invention, there is also a welding parameter dynamic optimization module: The input layer of the welding parameter optimization model based on the fuzzy neural network receives 12 input parameters, including the material of the connecting piece (copper, aluminum, etc., converted into digital signals through coding), thickness, width, length, welding position (flat welding, vertical welding, etc., also coded), and real-time monitored environmental temperature and humidity, electromagnetic interference intensity, current welding current, voltage, temperature, speed, and other parameters. In the fuzzy inference system, each input parameter is divided into different fuzzy sets according to its physical meaning and value range. For example, the temperature is divided into three fuzzy sets of "low", "medium", and "high", and the membership function (using a Gaussian membership function) is used to determine the degree to which the parameter belongs to each fuzzy set.

[0050] The neural network part adopts a three-layer structure, with 12 nodes in the input layer, 20 nodes in the hidden layer, and 3 nodes in the output layer (corresponding to welding current, welding time, and welding speed respectively). The network is trained using the backpropagation algorithm, and the network parameters are optimized through the genetic simulated annealing algorithm. During the training process, according to the formula Control annealing process, cooling coefficient The initial temperature T0 is set to 0.95, and the initial temperature T0 is set to 100. Using welding quality (assessed through indicators such as weld strength and appearance), welding efficiency (number of welds completed per unit time), and energy consumption (electricity consumed during welding) as output objectives, the NSGA-II algorithm is used to generate a set of Pareto optimal solutions through a multi-objective optimization strategy. In actual application, operators can select the appropriate welding parameter combination from the Pareto optimal solution set based on production requirements, and the system automatically sends the parameters to the welding equipment for adjustment.

[0051] In the present invention, the laser displacement sensor in the multi-dimensional data acquisition module adopts dual-wavelength interferometry measurement technology, which uses two lasers with different wavelengths for measurement, and uses the formula (in To measure distance changes, and For two laser wavelengths, The sensor uses a self-adaptive measurement range adjustment strategy, automatically adjusting the measurement range based on the approximate shape and size of the connector, to improve data acquisition efficiency and accuracy.

[0052] In this invention, the data processing and analysis module utilizes image enhancement technology based on a generative adversarial network (GAN) to process visual images. A generator network generates high-quality enhanced images based on low-quality input images, while a discriminator network determines the authenticity of the generated images. Through adversarial training, the quality of the generated images is continuously improved. Furthermore, a fast feature extraction algorithm based on geometric feature matching, combined with geometrically invariant image features (such as outline, angle, and scale), quickly and accurately locates key feature points of connectors, improving the speed and accuracy of feature recognition.

[0053] In this invention, the multi-dimensional motion platform of the adaptive positioning and adjustment module utilizes magnetic levitation support technology to reduce mechanical friction and vibration, improving motion smoothness and precision. Furthermore, the motion platform is equipped with a high-precision force sensor that monitors the forces acting between the connector and the welding tool in real time. Using a force feedback control algorithm, the robot arm's motion and posture are adjusted to ensure uniform force distribution during welding, preventing deformation and positioning errors caused by uneven force distribution.

[0054] In the present invention, the welding process monitoring module adopts a temperature monitoring system based on fiber optic sensing technology. Fiber optic sensors are distributed around the welding area, capable of monitoring the temperature changes during the welding process in real time and with high precision, and the precision can reach ±0.1°C. By establishing a temperature field model and analyzing the influence of the temperature distribution on the welding quality and the deformation of the connecting parts, using the formula (where T is the temperature, q is the heat source intensity, k is the thermal conductivity, is the thermal diffusivity) to predict the temperature change trend, providing a basis for the real-time adjustment of the welding process.

[0055] In the present invention, the electromagnetic environment perception and compensation module adopts a distributed electromagnetic sensor network, and realizes data sharing and collaborative work among the sensors through wireless communication technology. At the same time, the system adopts an electromagnetic interference prediction algorithm based on machine learning, and according to factors such as historical electromagnetic environment data and current meteorological conditions, predicts the occurrence probability and intensity of electromagnetic interference in advance, providing a more sufficient preparation time for electromagnetic interference compensation.

[0056] In the present invention, the welding parameter dynamic optimization module adopts a multi-objective optimization strategy, comprehensively considering multiple objectives such as welding quality, welding efficiency, and energy consumption. By introducing the Pareto optimal theory, a set of non-dominated solutions is generated, and the user can select the most suitable combination of welding parameters according to actual needs. At the same time, the system adopts an online learning mechanism, and continuously adjusts the parameters of the optimization model according to the real-time welding results, improving the accuracy and adaptability of the optimization.

[0057] In the present invention, the system further includes a remote collaboration and intelligent diagnosis module: realizing remote collaboration and data sharing among multiple welding positioning systems through an industrial Internet platform. Different systems can exchange welding data, process parameters, and fault information in real time, and jointly solve complex welding problems. At the same time, the system adopts an intelligent diagnosis algorithm based on deep learning to monitor the operating state of the system in real time and diagnose faults. By analyzing sensor data and system logs, it can quickly and accurately locate the cause of the fault and provide corresponding repair suggestions and solutions. Operators can view the system status, receive fault alarms, and perform remote operations through a mobile terminal or a remote monitoring center, improving the operation and maintenance efficiency and reliability of the system.

[0058] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A multi-dimensional adaptive conductive connector welding positioning system, characterized in that, It includes the following modules: Multi-dimensional data acquisition module: Deploy at least 5 laser displacement sensors on the welding workbench, and use the triangulation optimization algorithm , where d is the measurement distance and L is the sensor baseline length, is the fixed emission angle, is the reflected light reception angle. Set a Rogowski coil current sensor in the welding circuit to monitor current changes. The temperature and humidity sensor uses an integrated sensor. All sensor data is transmitted to the data processing module; Data processing and analysis module: Use the multi-scale feature fusion network MSFF-Net based on deep learning to process data. For laser displacement data, an adaptive wavelet threshold denoising algorithm is adopted, and the formula is , is the threshold of the i-th scale, is the noise standard deviation, is the number of data points at this scale; Adaptive positioning adjustment module: The multi-dimensional motion platform uses an eight-axis linkage robotic arm. The MPC algorithm pre-plans the motion trajectory of the robotic arm and optimizes the control input through the formula where \(u\) is the control input sequence, is the predicted output, is the reference output, \(Q\) and \(R\) are weight matrices. The robotic arm adopts a flexible joint design and adaptively adjusts the joint stiffness according to the shape and force condition of the connecting piece; Welding process monitoring module: It monitors the welding process by integrating multi-source data and adopts a state estimation method based on Kalman filter. The formula is , is the updated state estimation, is the predicted state estimation, is the Kalman gain, is the measured value, H is the measurement matrix. At the same time, a deep reinforcement learning algorithm is used to evaluate and optimize the welding process in real time; Intelligent Decision-making and Feedback Module: An intelligent decision-making system based on a knowledge graph is used. Knowledge reasoning and mining are carried out through graph neural networks. When there are deviations or faults in the system, the intelligent decision-making system analyzes the reasons and generates processing strategies. At the same time, the system adopts an active learning mechanism to update the knowledge graph and transmits the decision results in a timely manner through a real-time feedback mechanism.

2. The multi-dimensional adaptive conductive connector welding positioning system according to claim 1, wherein It also includes an electromagnetic environment perception and compensation module. Electromagnetic sensors are arranged around the system to monitor the intensity, frequency, and direction of the electromagnetic environment in real time. By establishing an electromagnetic interference model, the impact of electromagnetic interference on sensor data acquisition and system control is analyzed, and the formula is used for electromagnetic interference compensation, where is the compensated electric field strength, is the original electric field strength, is the compensation coefficient, is the electric field strength of each interference source, is the frequency of the interference source, is the phase of the interference source. At the same time, the system adopts an adaptive electromagnetic shielding strategy to dynamically adjust the parameters and structure of the shielding layer according to the changes in the electromagnetic environment.

3. The multi-dimensional adaptive conductive connector welding positioning system according to claim 1, characterized in that It also includes a welding parameter dynamic optimization module, which establishes a welding parameter optimization model based on a fuzzy neural network. The input parameters are fuzzified through a fuzzy inference system, and the learning ability of the neural network is used to automatically adjust the fuzzy rules and membership functions. The genetic simulated annealing algorithm is used to optimize and train the model, and the annealing process is controlled by the formula where is the temperature of the k-th generation, is the temperature reduction coefficient. During the welding process, the welding parameters are dynamically adjusted according to the real-time monitored data and the changes in the welding state.

4. The multi-dimensional adaptive conductive connector welding positioning system according to claim 1, wherein The laser displacement sensor in the multi-dimensional data acquisition module adopts the dual-wavelength interference measurement technology, measures through two lasers with different wavelengths, and uses the formula to detect tiny shape changes and defects on the surface of the connecting piece, where is the measured distance change, and are the two laser wavelengths, is the number of interference fringe changes. At the same time, the sensor adopts an adaptive measurement range adjustment strategy to automatically adjust the measurement range according to the general shape and size of the connecting piece.

5. The multi-dimensional adaptive conductive connector welding positioning system according to claim 1, characterized in that In the data processing and analysis module, for visual images, by establishing a dynamic physical model of the conductive connector and using the formula to evaluate the influence of the welding heat effect on positioning, where Q is the total heat effect, m is the mass of the connector, c is the specific heat capacity, is the temperature change, I is the welding current, R is the resistance, t is the welding time, is the influence coefficient of each environmental factor, is the quantization value of the corresponding environmental factor. The processing of visual images by this module adopts an image enhancement technology based on a generative adversarial network. The generator network generates high-quality enhanced images according to the input low-quality images, and the discriminator network judges the authenticity of the generated images. At the same time, a feature extraction algorithm based on geometric feature matching is adopted, combined with the geometric invariant features of the images, to locate the key feature points of the connector.

6. The multi-dimensional adaptive conductive connector welding positioning system according to claim 1, wherein The multi-dimensional motion platform of the adaptive positioning and adjustment module adopts magnetic levitation support technology to reduce mechanical friction and vibration. At the same time, the motion platform is equipped with force sensors to monitor the force between the connecting piece and the welding tool in real time. The movement and posture of the robotic arm are adjusted through a force feedback control algorithm to ensure uniform force on the connecting piece during the welding process.

7. The multi-dimensional adaptive conductive connector welding positioning system according to claim 1, wherein The welding process monitoring module adopts a temperature monitoring system based on fiber optic sensing technology. Fiber optic sensors are distributed around the welding area to monitor the temperature changes during the welding process. By establishing a temperature field model, the influence of temperature distribution on welding quality and the deformation of connecting parts is analyzed, and the temperature change trend is predicted using the formula where T is the temperature, q is the heat source intensity, k is the thermal conductivity, is the thermal diffusivity.

8. The multi-dimensional adaptive conductive connector welding positioning system according to claim 2, characterized in that, The electromagnetic environment perception and compensation module adopts a distributed electromagnetic sensor network to achieve data sharing and collaborative work among sensors through wireless communication technology. At the same time, the system adopts an electromagnetic interference prediction algorithm based on machine learning to predict the occurrence probability and intensity of electromagnetic interference in advance according to historical electromagnetic environment data and current meteorological condition factors.

9. The multi-dimensional adaptive conductive connector welding positioning system according to claim 3, wherein The welding parameter dynamic optimization module adopts a multi-objective optimization strategy. By introducing the Pareto optimality theory, a set of non-dominated solutions are generated, and the user selects the welding parameter combination according to actual needs. At the same time, the system adopts an online learning mechanism to adjust the parameters of the optimization model according to the real-time welding results.

10. The multi-dimensional adaptive conductive connector welding positioning system according to claim 1, characterized in that, The system also includes a remote collaboration and intelligent diagnosis module, which realizes remote collaboration and data sharing among multiple welding positioning systems through an industrial Internet platform. Welding data, process parameters, and fault information are exchanged in real time among different systems. At the same time, the system adopts an intelligent diagnosis algorithm based on deep learning to monitor the operation status of the system and diagnose faults in real time. By analyzing sensor data and system logs, the cause of the fault is located, and corresponding maintenance suggestions and solutions are provided. The operator can view the system status, receive fault alarms, and perform remote operations in real time through a mobile terminal or a remote monitoring center.

Images