A control method for the crystallization welding of dissimilar materials based on microwave energy distribution

Abstract

The present application provides a control method for the crystallization welding of dissimilar materials based on microwave energy distribution, including: obtaining a high-density temperature map of the workpiece to be welded through a fiber Bragg grating array, collecting real-time temperature values from the interface edge to the central region of the dissimilar materials of the workpiece to be welded to obtain an initial temperature distribution; calculating the temperature difference value between the edge and the central region, identifying the position of local overheating points by comparing the temperature gradient or setting a temperature difference threshold, and comparing the overheating point temperature with a preset material overheating threshold to determine whether there is an overheating region; if the temperature difference value exceeds the temperature difference threshold or there is a local overheating region, extracting the spatial coordinates of the low-temperature region and the high-temperature region from the high-density temperature map to identify the target region of the energy distribution that needs to be adjusted; extracting the temperature distribution from the updated energy distribution map, analyzing the temperature gradient to identify the heat concentration region, and calculating the target stress distribution state in combination with the heat stress concentration region.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to a method for controlling crystallization welding of heterogeneous materials based on microwave energy distribution. Background Art

[0002] Microwave energy regulation technology is the key to achieving efficient and non-contact heating. Microwave energy regulation technology occupies a key position in modern industrial manufacturing, especially in the field of welding of dissimilar materials, and has attracted much attention because of its ability to achieve efficient and non-contact heating. Microwave welding is an important process that uses this energy to connect materials, and its welding quality directly depends on the accuracy and uniformity of energy application. With the growing demand for high-precision and high-reliability welding in the aerospace, automobile manufacturing and new energy industries, how to achieve uniform temperature distribution and stable bonding quality in complex material systems has become a core issue to promote technological progress. Traditional microwave heating methods usually rely on a single feed or fixed phase configuration, which is difficult to adapt to large-size or material combinations with significant differences in thermal conductivity, such as welding scenes of glass panels and metal frames. When facing complex interfaces, these methods often cause local overheating or insufficient heating of the edge area due to uneven energy distribution, which in turn causes problems such as thermal stress concentration and decreased bonding strength. At this stage, although multi-point temperature monitoring technology has developed, how to combine high-density temperature data with dynamic regulation of microwave feeds is still a bottleneck that needs to be broken through. Specifically, the core challenges faced by current technologies are concentrated in the following aspects: First, the ability to acquire and process high-density temperature maps in real time is insufficient, which makes it difficult for the control system to accurately identify interface temperature differences or local hot spots; in addition, how to change the phase of the microwave feed signal fed into each microwave feed source by adjusting the phase shifter, and then change the spatial distribution form of the microwave energy field synthesized by multiple feed sources, so as to achieve precise control of the energy distribution in the welding area, is a technical difficulty that needs to be solved urgently. Secondly, the phase and power coordinated adjustment mechanism of the multi-feed microwave system is not mature enough, and it is difficult to dynamically optimize the energy distribution according to temperature feedback; finally, for heterogeneous materials with large differences in thermal conductivity, there is a lack of effective closed-loop control strategies to balance the overall temperature uniformity and local thermal stress risks. These unresolved technical factors directly lead to problems such as uncontrolled temperature distribution and unstable bonding quality during welding. Therefore, how to achieve precise control of microwave energy to the low temperature zone by dynamically adjusting the relative phase and power output of multiple feed sources based on the real-time feedback of the multi-point fiber grating temperature sensor array has become a key issue in improving the welding quality of heterogeneous materials. Summary of the invention

[0003] The present invention provides a method for controlling crystallization welding of dissimilar materials based on microwave energy distribution, which mainly includes:

[0004] Obtain the high-density temperature map of the workpiece to be welded through the fiber Bragg grating array, collect the real-time temperature values from the interface edge to the central area of the dissimilar materials of the workpiece to be welded, and obtain the initial temperature distribution;

[0005] Calculate the temperature difference value between the edge and the central area, identify the position of the local overheating point by comparing the temperature gradient or setting the temperature difference threshold, and compare the overheating point temperature with the preset material overheating threshold to determine whether there is an overheating area;

[0006] If the temperature difference value exceeds the temperature difference threshold or there is a local overheating area, extract the spatial coordinates of the low-temperature area and the high-temperature area from the high-density temperature map, and identify the target area of the energy distribution that needs to be adjusted;

[0007] Obtain the workpiece structure, material properties and the position of the feed source, establish the spatial distribution model of the microwave feed source signal energy field through electromagnetic simulation, calculate the power output and phase state of different feed sources in combination with the target area coordinates, and obtain the energy distribution deviation;

[0008] Change the phase of the signals fed into each microwave feed source through the phase shifter, adjust the control parameters of the phase shifter corresponding to each feed source by using the energy distribution deviation, increase the power output of the corresponding feed source through the spatial coordinates of the low-temperature area, and at the same time reduce the output intensity of the feed source in the high-temperature area to determine the target phase and the target power configuration;

[0009] According to the target phase and the target power configuration, perform simulation calculations using the spatial distribution model, obtain the predicted result of the updated microwave energy spatial distribution, and judge whether the updated microwave energy spatial distribution meets the temperature uniformity index to obtain the updated energy distribution map;

[0010] Extract the temperature distribution from the updated energy distribution map, analyze the temperature gradient to identify the thermally concentrated area, and calculate the target stress distribution state in combination with the thermally stressed concentrated area.

[0011] Further, in step S101, a high-density temperature map of the workpiece to be welded is obtained through a fiber grating array, and the real-time temperature values in the area from the interface edge to the center of the dissimilar materials of the workpiece to be welded are collected to obtain the initial temperature distribution, including: according to the preset fiber Bragg grating array layout scheme, the fiber Bragg grating array is arranged on the surface of the workpiece to be welded according to the density rule of at least one grating per square centimeter, the array sampling frequency is set to the number of samples per second according to the grating sensitivity coefficient, and the original wavelength value is obtained by collecting the central wavelength data from the fiber Bragg grating array. The collected original wavelength data is preprocessed by Gaussian filtering to eliminate the random fluctuation error, and the filtered wavelength data is demodulated by a fiber grating demodulation module, and the real-time temperature value is calculated according to the linear calibration curve of temperature and wavelength. A rectangular coordinate system temperature distribution grid is established for the obtained temperature data, and the Sobel edge detection operator is used to identify the interface edge contour line of the dissimilar materials, and the temperature contour line is drawn from the edge contour line to the central area at equal intervals. A cubic spline surface function is constructed based on the temperature contour line data to obtain a continuous temperature distribution surface equation, and the temperature gradient vector field from the edge to the central area is calculated. The bilinear interpolation algorithm is used to improve the spatial resolution of the temperature gradient vector field, and a high-density temperature distribution map is drawn according to the interpolated temperature gradient data to generate the initial distribution state of the surface temperature field of the workpiece to be welded.

[0012] Further, in step S102, calculate the temperature difference value between the edge and the central region, identify the local overheating point position by comparing the temperature gradient or setting a temperature difference threshold, and compare the overheating point temperature with the preset material overheating threshold to determine whether there is an overheating region, including: for the workpiece surface temperature monitoring point group, use the distance threshold clustering method to divide the edge region and the central region, determine the central region range inside the closed curve formed by the temperature monitoring points in the edge region, calculate the absolute value of the temperature difference between the edge and the central region from the temperature data collected from all monitoring points to obtain the regional temperature difference data. Construct a temperature difference matrix according to the regional temperature difference data, use the least squares method to calculate the temperature gradient vector field, segment the vector field through the set temperature gradient threshold, and extract the local overheating point position coordinates with a temperature gradient greater than the threshold from the segmented region. For the obtained overheating point position coordinates, extract the real-time temperature measurement value at this position, use a sliding mean filter with a window size of 5 to smooth the real-time temperature data, and judge the smoothed temperature value according to the overheating temperature threshold set according to the material melting point curve. Extract the temperature field data of the region around the overheating point from the temperature monitoring data matrix, use the Gaussian weighted average method to calculate the temperature distribution in this region, and determine the set of overheating region boundary points based on the temperature distribution data. Construct an overheating range contour curve based on the set of overheating region boundary points, use the curve fitting method to generate the overheating region boundary equation, and obtain the actual range of the overheating region by calculating the area enclosed by the boundary curve. For the temperature field data within the overheating region, use the finite difference method to calculate the temperature gradient field, extract the maximum gradient direction from the temperature gradient field data, and perform feature analysis on the overheating region based on the gradient direction.

[0013] Further, in step S103, if the temperature difference value exceeds the temperature difference threshold or there is a local overheating area, the spatial coordinates of the low-temperature area and the high-temperature area are extracted from the high-density temperature map, and the target area of the energy distribution that needs to be adjusted is identified, including: extracting the real-time temperature values of all monitoring points from the high-density temperature map, using a dynamic threshold segmentation method based on the temperature mean to stratify the temperature data, comparing the temperature difference values between adjacent levels according to the preset temperature difference threshold, and if the temperature difference value exceeds the temperature difference threshold, it is marked as a temperature difference abnormal area. For the marked temperature difference abnormal area, a temperature similarity threshold is set as the region growing criterion, and the region is expanded with the temperature abnormal point as the seed point. When the temperature difference between adjacent points is less than the similarity threshold, they are classified into the same region to obtain the local overheating area. According to the boundary point coordinates of the local overheating area, the parameters of the minimum circumscribed rectangle of the area are calculated, including the center point coordinates, length, width, and tilt angle of the rectangle, to obtain the spatial range of the overheating area. A temperature contour map is constructed using the temperature gradient calculation method, and the extreme points in the temperature field are extracted based on the local maximum suppression method. The high-temperature area and the low-temperature area are divided by setting a high-temperature threshold and a low-temperature threshold. According to the spatial distribution characteristics of the high-temperature area and the low-temperature area, the recursive quadtree method is used to divide the adjustment target area into grids, and the grid size is set according to the temperature uniformity within the grid. For the divided grid cells, the density-based spatial clustering method is used to group the grids, and the energy distribution adjustment area is determined by calculating the temperature distribution characteristic values of the grid groups.

[0014] Further, step S104, obtain the workpiece structure, material properties and feed source position, establish a microwave feed source signal energy field spatial distribution model through electromagnetic simulation, calculate the power output and phase state of different feed sources in combination with the target area coordinates, and obtain the energy distribution deviation, including: establishing a hexahedral grid division according to the three-dimensional geometric structure data of the workpiece, dividing the workpiece according to the grid size of one tenth of the minimum side length, reading the dielectric constant, magnetic permeability, and loss factor values of each part of the workpiece from the material parameter library, and determining the spacing and coordinates of each microwave feed source based on the feed source layout specification. Use a hexagonal grid to re-divide the workpiece surface, establish a spatial reference system through the grid unit vertex coordinates, assign values to the grid units according to the material parameters, and obtain the electromagnetic characteristic distribution data of the workpiece surface. Based on the finite difference time domain equations, set the perfect matching layer boundary conditions, iteratively calculate the microwave field distribution on the workpiece surface, extract the electric field intensity component and the magnetic field intensity component from the field distribution data, and establish the microwave field energy distribution function. A uniform sampling grid is established for the target area, and the bicubic spline interpolation method is used to reconstruct the field intensity data at the sampling points. The sampling point density is optimized by the minimum variance criterion to obtain high-precision field intensity distribution data. The feed source excitation equation group is constructed according to the field intensity distribution data, and the conjugate gradient method is used to solve the equation group to obtain the power output parameters and phase control parameters of each feed source, and establish the mapping relationship between the feed source parameters and the field intensity distribution. The energy deviation value of each sampling point is calculated by comparing with the standard field intensity distribution curve, and the energy distribution deviation function is obtained by the piecewise linear fitting method. The energy distribution deviation value of each region is obtained by numerical integration.

[0015] Further, in step S105, the phase of the signals fed into each microwave feed is changed by a phase shifter, and the control parameters of the phase shifters corresponding to each feed are adjusted by using the energy distribution deviation. The power output of the corresponding feed is increased through the spatial coordinates in the low-temperature area, and at the same time, the output intensity of the feed in the high-temperature area is reduced. Determining the target phase and target power configuration includes: establishing a phase-power response matrix according to the energy distribution deviation data, solving the phase optimization equation by using a genetic algorithm, using binary representation of the phase adjustment amount for chromosome coding, setting the fitness function as the root mean square error of the energy distribution deviation, and obtaining the phase adjustment parameters of each phase shifter. The power coupling relationship between each feed is calculated by using the partition summation method, the energy difference between the low-temperature area and the high-temperature area is extracted from the temperature field data, and the power adjustment increment of each feed is calculated according to the preset power adjustment step. For the phase adjustment parameters of the feeds in the low-temperature area, the phase compensation amount is calculated by using the orthogonal test method, the phase shifter parameters are corrected by the phase compensation amount, and the mapping relationship between the phase shifter parameters and the temperature field distribution is established. Based on the power adjustment increment of the feeds in the high-temperature area, the fitting function of the power transmission characteristic curve is set, and the power attenuation coefficient of the feeds in the high-temperature area is solved by using the linear programming method. A feed response equation set is established according to the corrected phase shifter parameters and power adjustment parameters, and the equation set is solved by using the least square method to obtain the target phase parameters and target power parameters of the feeds. A feed control sequence is constructed for the target phase parameters and target power parameters, and the phase adjustment curve and power adjustment curve are calculated by using the piecewise linear interpolation method to obtain the phase control amount and power control amount of the feeds.

[0016] Further, in step S106, according to the target phase and target power configuration, use the spatial distribution model for simulation calculation to obtain the updated predicted result of the microwave energy spatial distribution, and determine whether the updated microwave energy spatial distribution meets the temperature uniformity index, and obtain the updated energy distribution map, including: constructing an electromagnetic wave superposition equation according to the target phase and target power configuration, setting the boundary of the calculation region to adopt a perfectly matched layer absorbing boundary condition, using the spectral element method to solve the discretized wave equation, and extracting the electric field intensity distribution data from the wave field calculation result. Based on the electric field intensity distribution data, establish a unit volume energy density calculation function, set the initial grid size to one-tenth of the wavelength, and use the gradient adaptive method to encrypt the grid in the region where the energy density changes violently to obtain high-precision energy distribution data. Extract the energy field distribution curve on the workpiece surface from the energy distribution data, set the sampling point spacing to one-fifth of the wavelength, and use the cubic spline interpolation method to reconstruct the energy field data to generate a continuous energy distribution function. Calculate the spatial uniformity index for the reconstructed energy distribution function, set the temperature standard deviation threshold and the expected temperature rise range, and use the region segmentation method to evaluate the temperature field to obtain the temperature uniformity data of each region. Establish a heat conduction calculation grid according to the temperature uniformity data, set the convective heat transfer boundary condition on the workpiece surface, and use the implicit difference format to solve the heat conduction equation to obtain the steady-state temperature field distribution. Construct a temperature contour map for the steady-state temperature field distribution, set the contour interval to 5% of the temperature rise amount, use the region filling algorithm to generate a temperature distribution cloud map, and extract the updated energy distribution map from the cloud map data.

[0017] Further, in step S107, the temperature distribution is extracted from the updated energy distribution map, the temperature gradient is analyzed to identify the heat concentration region, and the target stress distribution state is calculated in combination with the thermal stress concentration region, including: extracting temperature distribution data from the updated energy distribution map, calculating the temperature gradient field by using the four-point central difference method, setting the temperature gradient threshold to 10% of the temperature rise amount, and constructing a heat concentration region map for the region where the temperature gradient exceeds the threshold. A hexagonal grid is established based on the heat concentration region map, the grid side length is set to one-fifth of the material thickness, the region growing method is used to group the grid cells, and the stress calculation boundary is determined from the grouping result. A stress equilibrium equation set is established for the divided grid cells, the material elastic modulus and Poisson's ratio parameters are set, and the stress components at the grid nodes are calculated by using the displacement method to obtain the initial stress distribution field. A stress feature data set is constructed according to the initial stress distribution field, including three types of characteristic parameters: stress amplitude, direction angle, and principal stress ratio. The characteristic data is preprocessed by using the normalization method. A convolutional neural network is established based on the preprocessed characteristic data, the convolutional kernel size and pooling parameters are set, and the network parameters are trained by using the stress distribution historical data to obtain the stress prediction model. The stress field distribution data is calculated from the stress prediction model, the stress field is reconstructed by using the bilinear interpolation method, and the combined stress distribution is calculated by setting the stress superposition coefficient to obtain the target stress state map.

[0018] The technical solution provided by the embodiment of the present invention may include the following beneficial effects:

[0019] The present invention discloses a method for controlling the crystal welding of dissimilar materials based on microwave energy distribution. The method obtains a high-density temperature map of the workpiece to be welded through a fiber Bragg grating array, identifies the positions of local overheating points, and extracts the spatial coordinates of the low-temperature region and the high-temperature region. Combining the workpiece structure, material properties, and the position of the feed source, a spatial distribution model of the microwave feed source signal energy field is established, and the power output and phase state of different feed sources are calculated. By adjusting the control parameters of each feed source through a phase shifter, increasing the power of the feed source in the low-temperature region and reducing the intensity of the feed source in the high-temperature region, the target phase and power configuration are achieved. Using the updated energy distribution map, the temperature gradient is analyzed to identify the heat concentration region, and the target stress distribution state is calculated in combination with the thermal stress concentration region. The present invention can effectively solve the problem of uneven temperature in the welding process of dissimilar materials, optimize the microwave energy distribution, and improve the welding quality. Description of the Drawings

[0020] Figure 1 It is a flowchart of a method for controlling the crystal welding of dissimilar materials based on microwave energy distribution according to the present invention.

[0021] Figure 2 It is a schematic diagram of a method for controlling the crystal welding of dissimilar materials based on microwave energy distribution according to the present invention. Detailed Embodiment

[0022] To further understand the content of the present invention, the present invention will be described in detail in combination with the accompanying drawings and embodiments. The following further elaborates on the present application in conjunction with the drawings and embodiments. It can be understood that the specific embodiments described herein are only used to explain the relevant invention and do not limit the invention. Additionally, it should be noted that for the convenience of description, only the parts related to the invention are shown in the drawings.

[0023] It can be understood that the method for controlling low-temperature welding of dissimilar materials based on microwave energy distribution in the present invention is used for low-temperature welding of the same or dissimilar materials by microwave special welding, which includes the following steps: First, clean and prepare the materials to be welded to ensure the surface is clean and pollution-free; Second, use a manipulator to transfer the materials to a designated position for heating, and then the welding head descends to a predetermined position above the materials; Third, set parameters such as the power, frequency, welding time, and temperature of the microwave according to the material characteristics and welding requirements; Fourth, start the microwave source for welding. After welding is completed, the welding head rises, and at the same time, use the temperature control system and data acquisition system to monitor the welding process to ensure the welding quality; Finally, after reaching the required welding time, turn off the microwave source, and the welding process is completed. The temperature and time of microwave low-temperature welding depend on the material characteristics and welding requirements. Generally, the welding temperature is lower than the melting point of the material, in the range of 200 - 300 °C, to avoid damaging the material. The welding time is usually between a few seconds and a few minutes.

[0024] Such as Figure 1 , a method for controlling the crystallization welding of dissimilar materials based on microwave energy distribution in this embodiment specifically may include:

[0025] S101 Collect the surface temperature data of the workpiece to be welded through a fiber Bragg grating array to generate an initial temperature distribution map.

[0026] In the embodiment of the present invention, a built-in welding control function can be started through the system interface or external instructions. The specific start method can be set by technicians according to requirements, such as operating through a touch panel or triggering by a remote signal. When the welding function is started, first confirm the workpiece to be welded and its material properties, and the density of the fiber Bragg grating array can be increased to at least two gratings per square centimeter at the key areas (such as the interface) at the interface of copper-aluminum dissimilar materials. During the temperature acquisition process, in addition to monitoring the welding temperature, attention should also be paid to the influence of environmental humidity and atmosphere on the welding quality, and an inert gas protection (such as argon) is introduced if necessary to prevent oxidation. For the characteristics of copper-aluminum materials, the preset overheat temperature thresholds should be set to approximately 1083 °C for copper and approximately 660 °C for aluminum respectively, and a certain safety margin should be considered on this basis.

[0027] Such as Figure 2, S1011 deploys a fiber Bragg grating array on the surface of the workpiece according to a preset layout plan, and collects the central wavelength data at a density of at least one grating per square centimeter.

[0028] In an embodiment of the present invention, the fiber Bragg grating array is arranged in a matrix to ensure coverage of key areas on the workpiece surface. The sensitivity coefficient of each grating is about 10 picometers per degree Celsius, and the sampling frequency is set to 100 times per second to capture real-time temperature changes. The typical value of the collected wavelength data is around 1550 nanometers, with a fluctuation range of about 1 nanometer, which is sufficient to reflect the temperature field change on the workpiece surface.

[0029] S1012 performs Gaussian filtering on the wavelength data, converts it into temperature values through a demodulation module after removing noise, and constructs a temperature distribution grid.

[0030] In an embodiment of the present invention, a Gaussian filter with a window size of 5 and a standard deviation of 0.8 is used to preprocess the wavelength data to remove random fluctuations. The filtered data is processed by a demodulation module and converted into temperature values based on the temperature-wavelength calibration curve, with an accuracy of 0.01 degree Celsius. Taking a corner of the workpiece surface as the origin, a coordinate system with a grid spacing of 1 millimeter is established for subsequent distribution calculation.

[0031] S1013 identifies the interface contour of dissimilar materials, draws isotherms from the contour to the central region, and generates a high-density temperature map.

[0032] In an embodiment of the present invention, the Sobel operator is used to analyze the temperature gradient to identify the material interface. When the gradient exceeds 50 degrees Celsius per centimeter, it is determined as the interface position. Isotherms are drawn every 5 degrees Celsius from the interface to the center, and a continuous temperature distribution equation is generated by fitting a cubic spline surface based on the isotherms. Further, bilinear interpolation is used to increase the grid resolution to 0.1 millimeter to draw a fine temperature map.

[0033] In an embodiment of the present invention, high-density temperature data is collected through a fiber Bragg grating array, which can accurately capture the temperature change on the workpiece surface from the edge to the center. Compared with traditional point monitoring, the high-resolution map generated by this embodiment of the present invention can clearly reflect the interface temperature difference and local hot spots.

[0034] S102 analyzes the temperature data on the workpiece surface, calculates the temperature difference between the edge and the central region, and identifies and locates local overheating points and overheating regions.

[0035] In an embodiment of the present invention, after the initial welding is completed, the resistance test of the joint is performed to ensure that the resistance is ≤0.06mΩ. If the resistance is found to exceed the standard, it is necessary to adjust the welding parameters (such as power, time) and re-weld. Using a high-density temperature map, the temperature change trend from the edge to the center of the workpiece surface is accurately analyzed. By calculating the regional temperature difference and combining it with gradient analysis, potential local hot spots are identified to ensure that abnormal temperature areas are discovered in time during the welding process. The temperature gradient threshold can be set to 50°C / cm, and areas exceeding this threshold are focused on. Compared with the traditional single monitoring method, the embodiment of the present invention can more accurately locate the overheating risk point and provide a reliable basis for subsequent energy regulation.

[0036] S1021 Based on the temperature monitoring point data, a clustering algorithm is used to divide the edge and center areas, calculate the temperature difference between the areas, and generate a temperature difference data matrix.

[0037] In an embodiment of the present invention, the coordinates of the monitoring points are extracted from the temperature data collected by the fiber grating array, and the workpiece surface is divided into an edge area and a central area by a distance-based clustering method. The maximum distance threshold between the monitoring points is set to 5 mm, and when the distance between two points is less than this value, they are classified as the same area. The edge area is defined by a closed contour close to the boundary point of the workpiece, and the interior is the central area. The average temperature difference between the monitoring points in the two areas is calculated. For example, for a rectangular workpiece with a side length of 100 mm, the temperature in the edge area is about 300 degrees Celsius, and the temperature in the central area is 500 degrees Celsius. A temperature difference data matrix is generated, and the temperature difference value of each grid point is recorded.

[0038] S1022 Use the least squares method to fit the temperature difference data matrix, construct the temperature gradient vector field, and extract the coordinates of the local hotspots.

[0039] In an embodiment of the present invention, the workpiece surface is divided into a grid with a resolution of 1 mm, and the temperature gradient vector field is calculated based on the temperature difference data matrix using the least squares method. The direction of the gradient vector indicates the path where the temperature changes most dramatically, and its modulus reflects the rate of change. The gradient threshold is set to 50 degrees Celsius per millimeter. When the gradient of a grid point exceeds this value, it is determined to be a local hot spot. The coordinates of the hot spot are recorded. For example, hot spots with coordinates of 45 mm, 48 mm, etc. are detected to provide a spatial reference for analysis.

[0040] S1023 performs smoothing on the hot spot temperature data, determines whether it exceeds the material overheating threshold, and determines the overheating area range.

[0041] In an embodiment of the present invention, for each overheat point, its real-time temperature value is extracted and smoothed using a mean filter with a sliding window size of 5 to eliminate measurement fluctuations. Taking the welding of aluminum and steel as an example, an overheat threshold of 600 degrees Celsius is set based on the melting point characteristics of aluminum. If the smoothed temperature exceeds this threshold, such as 620 degrees Celsius, it is marked as an overheat risk point. Further, temperature data within a range of 10 millimeters around the hot point is extracted, and the Gaussian weighted average method is used to calculate the regional temperature distribution, where the weight decays exponentially with distance, and the decay coefficient is set to 0.2 per millimeter, to determine the boundary points where the temperature is 80% lower than the threshold.

[0042] S1024 Fit the contour curve of the overheat area based on the boundary points, and calculate the area and gradient characteristics of the area.

[0043] In an embodiment of the present invention, using the boundary point data, the contour curve equation of the overheat area is generated by cubic spline interpolation. Calculate the area enclosed by the curve, such as approximately 75 square millimeters, to quantify the overheat range. The finite difference method is used to analyze the temperature gradient field within the overheat area, and the maximum gradient direction is extracted, which is usually consistent with the movement path of the welding heat source. For example, the gradient of a certain overheat point reaches 80 degrees Celsius per millimeter, which is much higher than the level of 30 degrees Celsius per millimeter around it, indicating that the heat is significantly concentrated. The overheat area is often elliptical, and the major axis is aligned with the welding direction, reflecting the heat conduction characteristics.

[0044] In an embodiment of the present invention, through clustering and gradient analysis, the temperature anomaly areas on the workpiece surface can be efficiently identified. Compared with the traditional method that relies on manually setting thresholds, the embodiment of the present invention combines Gaussian weighting and curve fitting to accurately quantify the overheat range and thermal characteristics, providing refined data support for dynamically adjusting the microwave energy, and significantly improving the uniformity of the welding temperature distribution and the process stability.

[0045] S103 Extract the spatial positions of the low-temperature area and the high-temperature area from the temperature distribution map, and determine the energy distribution area that needs to be adjusted.

[0046] In an embodiment of the present invention, based on the high-density temperature data, the temperature distribution characteristics of the workpiece surface are analyzed to accurately locate the temperature difference anomalies and overheat areas. The recursive quadtree algorithm can be used to divide the welding area into smaller grid units, and the size of each unit is about 1 millimeter side length or smaller to ensure that the temperature is uniform within each area, so as to ensure that the filler metal can uniformly fill the weld seam and increase the brazing rate to more than 85%. Through the dynamic threshold segmentation and region growing method, the key points of uneven energy distribution can be quickly identified, providing accurate spatial targets for microwave feeder control. At the same time, the energy input is dynamically adjusted according to the temperature monitoring data to avoid the generation of pore defects caused by local overheating leading to the boiling of the filler metal. Compared with the traditional static analysis method, the embodiment of the present invention significantly improves the recognition efficiency and accuracy of the temperature anomaly area.

[0047] S1031 Extract the temperature data of the monitoring points from the temperature map, generate temperature levels using dynamic threshold segmentation, and mark the regions with abnormal temperature differences.

[0048] In the embodiment of the present invention, a temperature map containing 100×100 monitoring points is processed, and the average temperature of all points is calculated, for example, 400 degrees Celsius, as the segmentation benchmark. The dynamic threshold is set to 1.2 times the average value, that is, 480 degrees Celsius, and the temperature data is divided into multiple levels. Compare the temperature differences between adjacent levels. When the temperature difference exceeds the preset threshold of 200 degrees Celsius, it is marked as an abnormal region. This method can effectively capture the temperature mutations in the heat-affected zones on both sides of the weld, providing basic data for subsequent analysis.

[0049] S1032 Using the abnormal points as seeds, perform region growing based on temperature similarity to determine the range of the local overheating region.

[0050] In the embodiment of the present invention, the temperature similarity threshold is set to 20 degrees Celsius, and the region is expanded starting from the points with abnormal temperature differences. When the temperature difference between adjacent points and the seed point is less than this threshold, they are classified into the same region. For example, in the scenario of welding aluminum and steel, three overheating regions are identified, and each region contains approximately 50 monitoring points. This method ensures that the boundary division of the overheating region conforms to the actual thermal distribution by comparing the temperature values point by point, improving the accuracy of region identification.

[0051] S1033 Calculate the minimum bounding rectangle of the overheating region to quantify its spatial distribution characteristics.

[0052] In the embodiment of the present invention, the sequence of boundary points of the overheating region is extracted, and the minimum bounding rectangle is fitted using the least squares method. Taking a certain overheating region as an example, the center of its rectangle is located at 45 mm and 55 mm on the surface of the workpiece, with a length of 30 mm, a width of 15 mm, and an inclination angle of 15 degrees. The rectangle parameters clearly describe the spatial range and direction of the overheating region, facilitating the locking of the target region during subsequent energy regulation and reducing unnecessary energy distribution deviation.

[0053] S1034 Construct a temperature isothermal line map, extract the extreme points, and divide the high-temperature region and the low-temperature region.

[0054] In the embodiment of the present invention, the temperature gradient of the monitoring points is calculated to generate an isothermal line map, with an interval set to 10 degrees Celsius. The maximum suppression algorithm with a 3×3 window is used to extract the temperature extreme points, and the high-temperature threshold is set to 600 degrees Celsius and the low-temperature threshold is set to 200 degrees Celsius. The high-temperature regions are mostly concentrated in the center of the weld, with the highest temperature reaching 800 degrees Celsius; the low-temperature regions are distributed in the periphery, with temperatures between 150 and 200 degrees Celsius. This division method intuitively reflects the spatial heterogeneity of the temperature field, providing a clear partition for energy optimization.

[0055] In the embodiment of the present invention, through grid division and clustering analysis, the energy adjustment area is further refined. The recursive quadtree algorithm is adopted, and the initial grid is set to have a side length of 10 mm. When the standard deviation of the temperature within the grid exceeds 50 degrees Celsius, it is subdivided into small grids with a side length of 2.5 mm. A dense grid group is formed near the high-temperature area, which contains about 20 units. The average temperature is between 620 and 750 degrees Celsius, and the standard deviation is controlled between 20 and 30 degrees Celsius. This adaptive division ensures the refinement of the regulation area, significantly improving the pertinence of energy distribution and the stability of the welding quality.

[0056] S104 According to the workpiece structure and material characteristics, combined with the position of the microwave feed, construct an energy field distribution model through electromagnetic simulation, and optimize the feed power and phase parameters.

[0057] In the embodiment of the present invention, using the three-dimensional geometric data of the workpiece and the electromagnetic characteristics of the material, accurately simulate the distribution of microwave energy on the surface of the workpiece. Through high-precision grid division and iterative calculation, an energy field model is generated to identify energy distribution deviations, providing a basis for dynamically adjusting the feed parameters. The melting process of the filler metal can also be accurately simulated through electromagnetic simulation to ensure that the filler metal is evenly filled and free of porosity defects. Compared with the traditional fixed feed configuration, the embodiment of the present invention significantly improves the uniformity of energy distribution and the stability of the welding process through refined modeling and optimization algorithms.

[0058] S1041 Generate hexahedral grids based on the workpiece geometric structure, assign material electromagnetic parameters, and determine the spatial layout of the feeds.

[0059] In the embodiment of the present invention, read the three-dimensional geometric data of the workpiece, such as an aluminum-steel composite workpiece with a length of 100 mm, a width of 50 mm, and a thickness of 10 mm. According to the criterion of one-tenth of the minimum side length, set the grid size to 1 mm and generate about 50,000 hexahedral elements. Extract the dielectric constant of 1.0, magnetic permeability of 1.0, and loss factor of 0.001 for aluminum, and the dielectric constant of 1.0, magnetic permeability of 1000, and loss factor of 0.01 for steel from the material database. According to the feed layout specification, evenly arrange four feeds on the surface of the workpiece with a spacing of 25 mm, and the coordinates are accurate to 0.1 mm. This way of grid division and parameter assignment ensures the spatial resolution of subsequent electromagnetic simulation and the accurate reflection of material characteristics.

[0060] S1042 Use the finite-difference time-domain method to calculate the microwave field distribution, extract the electric and magnetic field components, and construct an energy distribution function.

[0061] In the embodiment of the present invention, a hexagonal grid reconstruction is performed on the surface of the workpiece with a side length of 0.5 mm to form a high-resolution reference system. A 2-mm-wide transition zone is set at the interface of dissimilar materials, and the parameters are calculated by linear interpolation to simulate the gradual change of material properties. The finite-difference time-domain algorithm is applied, and an eight-layer perfectly matched layer boundary is set with an absorption coefficient of 0.333, and the electric and magnetic field components are calculated iteratively. The results show that the energy density in the central region of the weld is the highest, reaching 1000 W per square meter. Through vector synthesis, an energy distribution function is generated, clearly describing the spatial characteristics of the microwave field and providing a data basis for the optimization of the target area.

[0062] S1043 Interpolate and reconstruct the field strength data of the target area, optimize the sampling density, and generate a high-precision distribution model.

[0063] In the embodiment of the present invention, a uniform sampling grid with a side length of 10 mm is constructed in the target area, and the field strength data is reconstructed by bicubic spline interpolation, and the resolution is improved from 1 mm to 0.1 mm. The interpolation process optimizes the distribution of sampling points through the minimum variance criterion. In the weld area with a large field strength gradient, the sampling density is increased to 4 points per square millimeter. The standard deviation of the reconstructed field strength data is reduced by about 50%, significantly improving the model accuracy. This method effectively captures the subtle changes in the energy distribution and lays a foundation for subsequent parameter optimization.

[0064] In the embodiment of the present invention, according to the high-precision field strength data, an excitation equation set including the power and phase of four feeders is constructed. The conjugate gradient method is used to solve it, the convergence threshold is set to 0.001, and the iteration upper limit is 1000 times. The power outputs are 800 W, 600 W, 400 W, and 200 W respectively, and the phase differences are 0°, 90°, 180°, and 270°. Comparing the calculated field strength with the standard distribution, the deviation in the center of the weld is 5%, and the deviation in the heat-affected zone is 8%. A deviation function is generated by piecewise linear fitting, and the numerical integration shows that the total deviation in the weld area is 12% and that in the heat-affected zone is 15%. These deviation data provide a quantitative basis for the dynamic regulation of microwave energy, significantly improving the temperature uniformity and bonding quality in the welding area.

[0065] S105 Dynamically adjust the phase and power of the microwave feeder signal through a phase shifter, optimize the energy distribution according to the temperature field data, and balance the heat distribution between the low-temperature area and the high-temperature area.

[0066] In the embodiment of the present invention, using the energy deviation data, the phase and power output of each feeder are precisely adjusted to ensure that the microwave energy is preferentially distributed to the low-temperature area while reducing the energy input to the high-temperature area. This dynamic regulation mechanism effectively alleviates the temperature non-uniformity problem in the welding of dissimilar materials. Compared with the traditional fixed-parameter control, the embodiment of the present invention significantly improves the temperature field uniformity and the reliability of the welding quality through real-time feedback and optimization algorithms.

[0067] S1051 Construct a response matrix based on the energy distribution deviation, and optimize the phase parameters of the phase shifter using the genetic algorithm.

[0068] In the embodiment of the present invention, a phase and power response matrix is generated. The rows of the matrix cover the phase adjustment from 0 to 360 degrees with a step of 10 degrees, and the columns cover the power range from 0 to 100% with a step of 5%. The genetic algorithm is used to solve the phase optimization problem. The chromosome encodes the phase value in 16-bit binary. The population size is set to 100, the crossover probability is 0.8, and the mutation probability is 0.05. The fitness function is defined as the root mean square error of the energy deviation. Through 500 iterations, the optimal phase adjustment values of each phase shifter are obtained. This optimization method can quickly converge to the global optimal solution and ensure the accurate matching of energy distribution to the temperature demand.

[0069] S1052 Calculate the power coupling effect between the feeds, and determine the power adjustment increment based on the temperature field data.

[0070] In the embodiment of the present invention, the partition summation method is used to analyze the energy superposition effect between the feeds. Taking four feeds with a spacing of 25 mm as an example, the calculated power coupling coefficient is about 0.15, indicating that 15% of the energy overlaps between adjacent feeds. The energy difference between the average temperature of 200 degrees Celsius in the low-temperature area and 600 degrees Celsius in the high-temperature area is extracted from the temperature field. The power adjustment step is set to 10%, and the increment of each feed is calculated. For example, the power of the feed in the low-temperature area is increased by 10% to 20%, and the power of the feed in the high-temperature area is decreased by 5% to 15%. This method effectively quantifies the coupling effect and improves the pertinence of power distribution.

[0071] S1053 Optimize the phase compensation of the feeds in the low-temperature area through orthogonal experiments, and correct the parameters of the phase shifter.

[0072] In the embodiment of the present invention, for the feeds in the low-temperature area, an orthogonal experiment table is designed, including factors such as the initial phase value, adjustment step, and compensation coefficient. The influence of the phase compensation amount on the temperature uniformity is analyzed through experiments, and the compensation range is from 0 to 90 degrees. The results show that the optimal compensation combination improves the temperature field uniformity in the low-temperature area by about 30%. The parameters of the phase shifter are corrected according to the compensation amount, and the mapping relationship between the parameters and the temperature distribution is established to ensure the high consistency between the phase adjustment and the heat demand.

[0073] In the embodiments of the present invention, the feed power in the high-heat area is finely regulated. The power transmission characteristic curve is fitted by a cubic polynomial, and the power attenuation coefficient is solved by the linear programming method, with a range of 0.6 to 0.8. After attenuation, the energy density in the high-heat area is reduced by 20% to 40%, effectively avoiding the risk of overheating. A feed response equation set containing 16 non-linear equations is constructed, and the coefficient matrix fuses phase and power parameters. The least squares method is used for iterative solution, with a convergence threshold of 0.001 and an iteration upper limit of 1000 times. The target phases of the four feeds obtained by the solution are 0 degrees, 85 degrees, 170 degrees, and 255 degrees, and the power ratios are 90%, 70%, 50%, and 30%. A 60-second control sequence is generated, including phase and power curves adjusted once per second, with a phase step of 10 degrees and a power step of 5%. By adjusting the phase first and then the power, interference between parameters is avoided, ensuring the stability of the regulation process and the optimization of the welding effect.

[0074] S106 Use the spatial distribution model for simulation verification, predict the updated microwave energy distribution, and evaluate the temperature field uniformity to generate an optimized energy distribution map.

[0075] In the embodiments of the present invention, through high-precision electromagnetic and heat conduction simulations, it is verified whether the adjusted feed parameters can effectively balance the temperature distribution on the workpiece surface. The simulation results generate continuous energy and temperature distribution maps, providing an intuitive basis for welding quality evaluation. Compared with the traditional empirical adjustment method, the embodiments of the present invention ensure that the energy distribution precisely matches the heat demand through multi-level calculations and verifications, significantly reducing the risks of local overheating and thermal stress.

[0076] S1061 Based on the target phase and power, construct an electromagnetic wave superposition equation, and use the spectral element method to calculate the electric field intensity distribution.

[0077] In the embodiments of the present invention, according to the target phases and powers of the four feeds, with a working frequency of 2.45 GHz and a wavelength of about 122 mm, an electromagnetic wave superposition equation is constructed. Eight layers of perfectly matched layers are set at the calculation region boundary, with an absorption coefficient of 0.333 to reduce boundary reflection interference. The spectral element method is used for solution, and an eighth-order polynomial basis function is selected, with the field strength calculation accuracy reaching 0.1 V / m. The generated electric field intensity distribution clearly reflects the high-energy concentration area at the weld center, providing high-resolution data for subsequent energy density analysis.

[0078] S1062 Calculate the energy density through the electric field data, and use adaptive grid encryption to improve the accuracy in high-gradient regions.

[0079] In the embodiment of the present invention, the energy density per unit volume is calculated based on the product of the square of the electric field strength and the permittivity. The initial grid size is set to one-tenth of the wavelength, approximately 12 mm. For the region where the energy density gradient exceeds 100 W / (mm³·mm), the grid is refined to 3 mm through the gradient adaptive algorithm. After three iterations, the calculation accuracy in the local region is improved by approximately four times. This method effectively captures the drastic changes in the energy distribution near the weld, ensuring the generation of a high-precision energy field.

[0080] S1063 Extract the energy distribution curve on the workpiece surface and generate a continuous energy function by interpolation reconstruction.

[0081] In the embodiment of the present invention, the energy density data on the workpiece surface is collected at an interval of 24 mm to generate a grid of 20×10 sampling points. Cubic spline interpolation is used to reconstruct the data. In the weld region where the energy changes drastically, the node spacing is reduced to 12 mm. The reconstructed continuous energy distribution function accurately describes the gradual change characteristics of the energy from the weld center to the edge, providing a reliable input for subsequent heat conduction calculations.

[0082] In the embodiment of the present invention, the temperature uniformity of the energy distribution is evaluated. The standard deviation threshold is set to 20 °C, and the expected temperature rise is 200 to 600 °C. The quadtree algorithm is used to divide the temperature field into 16 sub-regions, and the temperature standard deviation of each region is calculated. The standard deviation at the weld center is approximately 15 °C, meeting the requirements; the standard deviation in the edge region reaches 25 °C, indicating that further optimization is needed. A heat conduction grid is constructed, with a heat transfer coefficient of 25 W / (m²·K) and an ambient temperature of 20 °C. The implicit difference method is used to solve the heat conduction equation, with a time step of 0.1 s and a spatial step of 1 mm. After 500 iterations, a steady-state temperature field is obtained. The highest temperature of 580 °C is located at the weld center. A temperature contour map is generated, with an interval of 20 °C, and 30 isotherms are plotted, filled with a blue-green-yellow-red gradient, forming an intuitive temperature cloud map that clearly shows the elliptical high-temperature region and the gradient distribution characteristics, providing a visual basis for optimizing energy regulation.

[0083] S107 Extract the temperature distribution data from the updated energy distribution map, analyze the heat concentration region and the stress distribution characteristics, and predict the target stress state in combination with the neural network algorithm.

[0084] In the embodiments of the present invention, by analyzing the temperature gradient and stress distribution, the thermal stress concentration regions that may cause cracks or deformations during the welding process are accurately identified. A deep learning model is used to predict the changes in the stress field, providing data support for optimizing the welding parameters. Combining the stress analysis results, a preliminary prediction of the fatigue resistance, corrosion resistance, etc. of the joint is carried out to ensure that the reliability requirements in practical applications are met. Special attention is paid to the durability performance of the joint under different environments to prevent corrosion problems during long-term use. Compared with traditional stress analysis methods, the embodiments of the present invention combine mesh division and neural network, greatly improving the accuracy and efficiency of stress prediction and effectively reducing the risk of welding defects.

[0085] S1071 Calculate the temperature gradient field from the energy distribution map and construct a thermal concentration region map.

[0086] In the embodiments of the present invention, the temperature data in the energy distribution map is extracted, and the gradient is calculated using the four-point central difference method with a grid spacing of 1 mm. The gradient threshold is set to 10% of the temperature rise amount. For example, when the temperature rise is 500 degrees Celsius, the threshold is 50 degrees Celsius per mm. The regions with higher gradients within the weld area are identified, such as hot spots with a gradient value of 75 degrees Celsius per mm, to generate a thermal concentration region map. This method can quickly locate the regions of rapid temperature change, providing a spatial basis for stress analysis.

[0087] S1072 Perform hexagonal mesh division on the thermal concentration region and use the region growing method to determine the stress calculation boundary.

[0088] In the embodiments of the present invention, taking one-fifth of the workpiece thickness as the standard, the side length of the hexagonal mesh is set. For example, for a 10-mm thick workpiece, a 2-mm side length mesh is used to generate approximately 500 elements, of which 150 are in the thermal concentration region. The region growing method is used for grouping, and adjacent elements are merged based on the similarity of temperature gradients to generate a continuous stress calculation boundary. Displacement constraints are applied at the boundary to ensure that the calculation results reflect the actual stress state. This division method takes into account both the calculation efficiency and the refinement of local regions.

[0089] S1073 Establish a stress equilibrium equation and calculate the initial stress distribution field.

[0090] In the embodiments of the present invention, a stress equilibrium equation set is constructed based on the grid elements, and the elastic modulus of aluminum is input as 70 GPa, Poisson's ratio as 0.33, and the elastic modulus of steel as 210 GPa, Poisson's ratio as 0.28. The stiffness matrix is generated using the displacement method, and the stress components are calculated after solving the nodal displacements. The results show that the maximum equivalent stress in the thermal concentration region is approximately 250 MPa. This method ensures the reliability of the initial stress field through accurate material parameters and boundary conditions.

[0091] S1074 Construct a stress feature dataset and train a convolutional neural network to generate a stress prediction model.

[0092] In an embodiment of the present invention, features are extracted from the initial stress field, including stress amplitude from 0 to 300 MPa, direction angle from 0 to 360 degrees, and principal stress ratio from 0.2 to 0.8. 1000 groups of sample data sets are generated, 80% of which are used for training and 20% for validation. The data is normalized to the range of 0 to 1. A five-layer convolutional neural network is designed, including three 3×3 convolutional layers and two fully connected layers, with a pooling window of 2×2. The network takes a 48×48 feature map as input and outputs the stress prediction value. The training uses a batch size of 32, a learning rate of 0.001, and 1000 iterations, reducing the prediction error to below 5%. This model can capture the non-linear features of the stress field and improve the prediction accuracy.

[0093] In an embodiment of the present invention, the stress prediction model is used to generate the stress field distribution. Bilinear interpolation is used to reconstruct the continuous stress field, and interpolation is calculated based on four neighboring points to ensure a smooth distribution. The thermal stress weight is set to 0.7 and the mechanical stress weight is set to 0.3 to calculate the combined stress. The results show that the maximum stress at the weld center is about 200 MPa, and the stress in the heat affected zone decreases from the center to 50 MPa outward, showing a gradient distribution. The stress curve shows non-linear changes at the transition between the weld and the heat affected zone, which is consistent with the actual welding stress evolution, providing an accurate stress state reference for subsequent process optimization.

[0094] The above-disclosed are only the preferred embodiments of the present invention. Of course, the scope of the rights of the present invention cannot be limited thereby. Those of ordinary skill in the art can understand all or part of the processes of implementing the above embodiments, and the equivalent changes made according to the claims of the present invention still fall within the scope covered by the invention.

Claims

1. A control method for the crystallization welding of dissimilar materials based on microwave energy distribution, characterized in that, The method includes: Obtaining a high-density temperature map of the workpiece to be welded through a fiber Bragg grating array, collecting real-time temperature values from the interface edge to the central region of the dissimilar materials of the workpiece to be welded, and obtaining the initial temperature distribution; Calculating the temperature difference value between the edge and the central region, identifying the position of local overheating points by comparing the temperature gradient or setting a temperature difference threshold, and comparing the overheating point temperature with a preset material overheating threshold to determine whether there is an overheating region; If the temperature difference value exceeds the temperature difference threshold or there is a local overheating region, extract the spatial coordinates of the low-temperature region and the high-temperature region from the high-density temperature map, and identify the target region of the energy distribution that needs to be adjusted; Obtain the workpiece structure, material properties, and feeder position, establish a spatial distribution model of the microwave feeder signal energy field through electromagnetic simulation, and calculate the power output and phase state of different feeders in combination with the target region coordinates to obtain the energy distribution deviation; Change the phase of the signals fed into each microwave feeder through a phase shifter, adjust the control parameters of the phase shifters corresponding to each feeder using the energy distribution deviation, increase the power output of the corresponding feeder through the spatial coordinates of the low-temperature region, and at the same time reduce the output intensity of the feeder in the high-temperature region to determine the target phase and target power configuration; According to the target phase and target power configuration, perform simulation calculations using the spatial distribution model to obtain the updated predicted result of the microwave energy spatial distribution, and determine whether the updated microwave energy spatial distribution meets the temperature uniformity index to obtain the updated energy distribution map; Extract the temperature distribution from the updated energy distribution map, analyze the temperature gradient to identify the thermally concentrated region, and calculate the target stress distribution state in combination with the thermally stressed concentrated region.

2. The method according to claim 1, wherein The obtaining a high-density temperature map of the workpiece to be welded through a fiber Bragg grating array, collecting real-time temperature values from the interface edge to the central region of the dissimilar materials of the workpiece to be welded, and obtaining the initial temperature distribution includes: Perform Gaussian filtering preprocessing on the central wavelength data generated on the workpiece surface to obtain the filtered wavelength data, and demodulate the filtered wavelength data through a fiber Bragg grating demodulation module to obtain temperature values; Use the Sobel edge detection operator to identify the interface edge contour line of the dissimilar materials for the temperature values, and draw temperature isotherms from the interface edge contour line to the central region; Construct a cubic spline surface function based on the temperature isotherms to obtain the temperature distribution surface equation, and use the bilinear interpolation algorithm to perform spatial resolution improvement processing on the temperature gradient vector field calculated by the temperature distribution surface equation to generate the initial distribution of the surface temperature field of the workpiece to be welded.

3. The method according to claim 1, wherein The calculating the temperature difference value between the edge and the central region, identifying the position of local overheating points by comparing the temperature gradient or setting a temperature difference threshold, and comparing the overheating point temperature with a preset material overheating threshold to determine whether there is an overheating region includes: Divide the edge region and the central region according to the workpiece surface temperature, calculate the temperature difference between the edge region and the central region, and obtain the temperature gradient vector field; Extract the position of local overheating points from the temperature gradient vector field through a preset temperature gradient threshold; According to the temperature data of the local overheating point position, the Gaussian weighted average method is used to calculate the temperature distribution in the area around the overheating point. The boundary equation of the overheating area is obtained by constructing the overheating range contour curve from the temperature distribution, and the actual range of the overheating area is obtained by calculating the area enclosed by the boundary curve.

4. The method according to claim 1, wherein If the temperature difference value exceeds the temperature difference threshold or there is a local overheating area, the spatial coordinates of the low-temperature area and the high-temperature area are extracted from the high-density temperature map, and the target area of the energy distribution that needs to be adjusted is identified, including: Obtain the real-time temperature values of the monitoring points in the high-density temperature map, calculate the temperature mean value according to the real-time temperature values, and obtain the temperature stratification data through dynamic threshold segmentation using the temperature mean value. Calculate the temperature difference value between adjacent levels for the temperature stratification data. If the temperature difference value exceeds the preset temperature difference threshold, the temperature difference abnormal area is obtained. Expand the area according to the abnormal point coordinates of the temperature difference abnormal area and the preset temperature similarity threshold, and classify the temperatures of adjacent points through the temperature similarity threshold to obtain the local overheating area. Calculate the minimum circumscribed rectangle parameters for the boundary point coordinates of the local overheating area, and construct a temperature isogram using the minimum circumscribed rectangle parameters. Extract the extreme points of the temperature field from the temperature isogram, divide the high-temperature area and the low-temperature area by setting the high-temperature threshold and the low-temperature threshold, perform grid division on the adjustment target area, group the grids using the spatial clustering method, and determine the target area of the energy distribution that needs to be adjusted by calculating the temperature distribution characteristic values of the grid groups.

5. The method according to claim 1, wherein Obtain the workpiece structure, material properties, and feed source positions, establish a spatial distribution model of the microwave feed source signal energy field through electromagnetic simulation, and calculate the power output and phase states of different feed sources in combination with the target area coordinates to obtain the energy distribution deviation, including: Establish a hexahedral mesh dissection according to the workpiece three-dimensional geometric structure data, assign electromagnetic characteristic parameters to the workpiece surface through the vertex coordinates of the grid cells, and obtain the electromagnetic characteristic distribution data of the workpiece surface. Calculate the microwave field distribution on the workpiece surface, extract the electric field strength component and the magnetic field strength component, and obtain the microwave field energy distribution function. Reconstruct the field strength data for the target area using the bicubic spline interpolation method, optimize the sampling point density through the minimum variance criterion, and obtain the field strength distribution data. Construct a feed source excitation equation set according to the field strength distribution data, solve the equation set using the conjugate gradient method, obtain the power output parameters and phase control parameters of the feed source, and calculate the energy deviation values of each sampling point by comparing with the standard field strength distribution curve.

6. The method according to claim 1, wherein Change the phase of the signals fed into each microwave feed source through the phase shifter, adjust the control parameters of the phase shifter corresponding to each feed source using the energy distribution deviation, increase the power output of the corresponding feed source through the spatial coordinates of the low-temperature area, and at the same time reduce the output intensity of the feed sources in the high-temperature area to determine the target phase and target power configuration, including: Establish a phase and power response matrix according to the energy distribution deviation. The phase and power response matrix uses the genetic algorithm to solve the phase optimization equation to obtain the phase adjustment parameters of the phase shifter. The power coupling relationship between feeds is calculated by the partition cumulative addition method. The energy difference between the low-temperature area and the high-temperature area is extracted from the temperature field data, and the power adjustment increment is obtained according to the power adjustment step size. For the phase adjustment parameters of the phase shifter, the orthogonal test method is used to calculate the phase compensation amount, and the corrected phase shifter parameters are obtained through the phase compensation amount. According to the corrected phase shifter parameters and the power adjustment increment, a feed response equation set is established, and the least squares method is used to solve the equation set to obtain the target phase parameters and target power parameters of the feeds.

7. The method according to claim 1, wherein According to the target phase and target power configuration, simulation calculation is carried out using the spatial distribution model to obtain the updated microwave energy spatial distribution prediction result, and it is judged whether the updated microwave energy spatial distribution meets the temperature uniformity index, and the updated energy distribution map is obtained, including: Construct an electromagnetic wave superposition equation according to the target phase and target power. The spectral element method is used to solve the discretized wave equation, and the electric field intensity distribution data is extracted from the wave field calculation result. According to the electric field intensity distribution data, a unit volume energy density calculation function is established, and the grid is encrypted in the area where the energy density changes violently to obtain the energy distribution data. The workpiece surface energy field distribution curve is extracted from the energy distribution data, and the energy field distribution curve is reconstructed by the cubic spline interpolation method. Calculate the spatial uniformity index of the reconstructed energy distribution function, set the temperature standard deviation threshold and the expected temperature rise range, and use the region segmentation method to evaluate the temperature field to obtain the temperature uniformity data of each region. Generate the updated energy distribution map according to the temperature uniformity data.

8. The method according to claim 1, wherein Extract the temperature distribution from the updated energy distribution map, analyze the temperature gradient to identify the thermal concentration area, and combine the thermal stress concentration area to calculate the target stress distribution state, including: Calculate the temperature gradient field data according to the energy distribution map, and use the four-point central difference method to construct a thermal concentration area map for the area where the temperature gradient exceeds the preset threshold. Establish hexagonal grid cells for the thermal concentration area map, and use the region growing method to group the grid cells to obtain the stress calculation boundary data. According to the stress calculation boundary data, establish a stress balance equation set, and use the displacement method to calculate the stress components at the grid cell nodes to obtain the initial stress distribution field. Calculate the stress field distribution data for the initial stress distribution field, and reconstruct the stress field based on the stress field distribution data to obtain the target stress state map.

Images