Flexible circuit board three-dimensional deformation prediction and compensation method based on deep learning

Abstract

The application provides a flexible circuit board three-dimensional deformation prediction and compensation method based on deep learning, relates to the technical field of circuit boards, and comprises the following steps: acquiring process parameters and actual deformation data to establish a relationship data set, constructing a variable-density grid unit to calculate a stress gradient matrix, forming a deformation prediction parameter matrix to perform three-dimensional deformation prediction, constructing a feedback compensation matrix after calculating a deformation deviation value, adjusting process parameters according to a compensation transmission link order, welding and processing a flexible circuit board, and realizing deformation compensation. The application can effectively improve the three-dimensional deformation prediction precision and welding quality of the flexible circuit board.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of circuit board, in particular to a flexible circuit board three-dimensional deformation prediction and compensation method based on deep learning. BACKGROUND

[0002] With the development of electronic devices towards light and thin, portable, flexible circuit board (FPC) is widely used in smart phones, wearable devices and other fields due to its lightweight, thin, bendable characteristics. In the production process of flexible circuit board, especially in the welding process, due to the mismatch of material thermal expansion coefficient, welding temperature gradient and mechanical stress and other factors, the flexible circuit board often deforms in three dimensions, including bending, twisting and warping, which seriously affects the assembly accuracy, electrical performance and reliability of electronic products.

[0003] The traditional flexible circuit board deformation control method mainly relies on empirical formula and simplified model for prediction, which is difficult to accurately reflect the actual deformation under complex process conditions. With the continuous improvement of manufacturing precision, the existing technology has obvious shortcomings. First, the traditional deformation prediction method usually adopts uniform grid division, which cannot be fine analyzed for the dense welding point area, resulting in insufficient deformation prediction accuracy in stress concentration area. Secondly, the existing compensation method often adopts static adjustment strategy, lacks consideration of the stress transmission relationship between multiple welding points, and the compensation effect is not ideal, especially under the influence of multiple welding points, which is prone to overcompensation or undercompensation phenomenon. In addition, most of the existing methods fail to establish an accurate mapping relationship between process parameters and deformation, and cannot adaptively adjust according to the dynamic changes in actual production, resulting in low accuracy and efficiency of deformation compensation.

[0004] With the continuous improvement of the assembly accuracy requirement of flexible electronic products, it is urgent to develop a method that can accurately predict the three-dimensional deformation of flexible circuit board and implement effective compensation to improve product quality and production efficiency. SUMMARY

[0005] The embodiment of the present application provides a flexible circuit board three-dimensional deformation prediction and compensation method based on deep learning, which can solve the problems in the prior art.

[0006] In a first aspect, the embodiment of the present application provides a flexible circuit board three-dimensional deformation prediction and compensation method based on deep learning, comprising:

[0007] Obtaining process parameters in the production process of flexible circuit board, and establishing a process parameter-deformation relationship data set according to the process parameters and the actual three-dimensional deformation data of the flexible circuit board;

[0008] A variable-density grid unit is constructed along the distribution direction of the solder joints of the flexible circuit board, a stress gradient matrix is calculated based on stress deformation characteristic data of the variable-density grid unit, and a deformation prediction parameter matrix is constructed based on the stress gradient matrix;

[0009] The process parameter-deformation relationship dataset is input into the deformation prediction parameter matrix, and three-dimensional deformation prediction data are calculated. The actual three-dimensional deformation data are compared with the three-dimensional deformation prediction data, and a deformation deviation value is calculated;

[0010] A feedback compensation matrix is constructed based on the deformation deviation value. Each compensation unit in the feedback compensation matrix forms a compensation transmission link in the stress transmission direction. The process parameters are sequentially adjusted according to the transmission characteristics of the compensation transmission link to obtain adjusted process parameters;

[0011] The flexible circuit board is welded according to the adjusted process parameters to compensate for the three-dimensional deformation of the flexible circuit board.

[0012] Obtain process parameters in the production process of the flexible circuit board, and establish a process parameter-deformation relationship dataset according to the process parameters and actual three-dimensional deformation data of the flexible circuit board, including:

[0013] Collect the process parameters of the flexible circuit board during welding, arrange the process parameters according to the execution time sequence of the welding process according to the collection time of the process parameters, and generate a process parameter time sequence;

[0014] Real-time collection of actual three-dimensional deformation data of the flexible circuit board under the action of the process parameter time sequence, calculation of deformation gradient of the actual three-dimensional deformation data at adjacent collection time;

[0015] Determine the deformation compensation direction according to the deformation gradient, construct a process parameter change matrix according to the process parameter difference of adjacent time sequence nodes in the process parameter time sequence, and calculate a deformation compensation coefficient based on the weight distribution of the temperature process parameter difference and the pressure process parameter difference in the process parameter change matrix;

[0016] Apply the deformation compensation direction to the deformation compensation coefficient to generate a deformation compensation amount of each time sequence node;

[0017] Construct a compensation adjustment curve in the temperature process parameter associated area and the pressure process parameter associated area respectively according to the deformation compensation amount, and perform regional progressive compensation on the process parameter time sequence according to the compensation adjustment curve to establish the process parameter-deformation relationship dataset.

[0018] constructing variable-density grid units along the distribution direction of the flexible circuit board solder joints, calculating a stress gradient matrix based on stress deformation characteristic data of the variable-density grid units, and constructing a deformation prediction parameter matrix based on the stress gradient matrix, including:

[0019] generating variable-density grid units along the distribution direction of the flexible circuit board solder joints, the variable-density grid units being set as micro-scale grid units in the solder joint distribution area and as macro-scale grid units in the non-solder joint area;

[0020] obtaining first stress deformation characteristic data in the micro-scale grid units, the first stress deformation characteristic data representing stress distribution and deformation trend in the solder joint area, and obtaining second stress deformation characteristic data in the macro-scale grid units, the second stress deformation characteristic data representing stress distribution and deformation trend in the non-solder joint area;

[0021] performing multi-scale deformation field reconstruction based on the grid density distribution of the first stress deformation characteristic data and the second stress deformation characteristic data to obtain a stress field reconstruction matrix;

[0022] calculating stress gradients of the variable-density grid units based on the stress field reconstruction matrix, and generating a stress gradient matrix according to the distribution positions of the variable-density grid units;

[0023] decoupling the stress field reconstruction matrix and the stress gradient matrix in terms of features, extracting stress deformation correlation features, and constructing a deformation prediction parameter matrix according to the stress deformation correlation features.

[0024] performing multi-scale deformation field reconstruction based on the grid density distribution of the first stress deformation characteristic data and the second stress deformation characteristic data to obtain a stress field reconstruction matrix, including:

[0025] dividing the micro-scale grid units and the macro-scale grid units into deformation reconstruction areas according to the size of the density gradient, calculating the density change rate of the deformation reconstruction areas, and constructing a reconstruction feature vector according to the density change rate;

[0026] generating a spiral reconstruction path based on the density distribution of the reconstruction feature vector, the spatial configuration of the spiral reconstruction path matching the density change process of the reconstruction feature vector in terms of geometric parameters, and performing density reconstruction mapping of the first stress deformation characteristic data and the second stress deformation characteristic data along the spiral reconstruction path to generate reconstruction enhanced data;

[0027] A reconstruction data transmission channel is established inside the morphing reconstruction region, a direction of the reconstruction data transmission channel is consistent with a density gradient direction of the reconstruction feature vector, and the reconstruction enhancement data is transmitted along the reconstruction data transmission channel to obtain transmitted reconstruction enhancement data;

[0028] A stress field reconstruction matrix is generated from the transmitted reconstruction enhancement data according to the density gradient direction of the reconstruction feature vector.

[0029] The process parameter-morphing relationship dataset is input into the morphing prediction parameter matrix, and three-dimensional morphing prediction data is calculated. The actual three-dimensional morphing data of the flexible circuit board is compared with the three-dimensional morphing prediction data, and a morphing deviation value is calculated, including:

[0030] The process parameter-morphing relationship dataset is input into the morphing prediction parameter matrix, and the process parameters are hierarchically quantized based on the stress transmission channel in the morphing prediction parameter matrix. A morphing prediction flow field is generated based on the quantized process parameters in the stress transmission channel.

[0031] The morphing prediction flow field is unfolded in three-dimensional space to obtain three-dimensional morphing prediction data, and a three-dimensional morphing prediction surface is generated based on the three-dimensional morphing prediction data.

[0032] A curvature constraint condition is constructed based on the morphing amount ratio between adjacent feature positions in the actual three-dimensional morphing data, and a three-dimensional morphing measured surface is generated by tensor completion of the actual three-dimensional morphing data. The curvature constraint condition is used to limit the spatial continuous features of the actual three-dimensional morphing data.

[0033] A continuous transition surface is constructed between the three-dimensional morphing prediction surface and the three-dimensional morphing measured surface, and a hyperboloid interference graph is constructed based on the hyperbolic transition features of the continuous transition surface.

[0034] The peak position and the valley position of the hyperboloid interference graph are extracted, and the spatial coordinate difference between the peak position and the valley position is calculated as a morphing deviation value.

[0035] A feedback compensation matrix is constructed based on the morphing deviation value. Each compensation unit in the feedback compensation matrix forms a compensation transmission link according to the stress transmission direction. The process parameters are sequentially adjusted according to the transmission characteristics of the compensation transmission link to obtain adjusted process parameters, including:

[0036] A morphing feature point is determined according to the distribution of the morphing deviation value, a morphing coordinate system is established based on the morphing feature point, and a stress transmission direction of the morphing feature point is calculated.

[0037] The compensation unit array is divided according to the stress transmission direction in the deformation coordinate system, the positional relationship between the deformation deviation value of each compensation unit and the deformation feature point is calculated, and the feedback compensation matrix is generated by arranging the deformation deviation value according to the position of the compensation unit;

[0038] The main transmission node is determined based on the stress distribution characteristics of the compensation units in the feedback compensation matrix, the compensation transmission link is established from the main transmission node, the compensation values in the compensation transmission link are spatially reconstructed based on the positional association relationship between the compensation units to generate a compensation distribution matrix;

[0039] The compensation values of the compensation units in the compensation unit array are adjusted based on the compensation values in the compensation distribution matrix to generate a continuous compensation field;

[0040] The spatial positional relationship of the compensation units in the compensation distribution matrix is converted into adjustment coordinates of the process parameters, the adjustment order of the process parameters in the adjustment coordinates is determined according to the transmission direction of the compensation transmission link, and the adjusted process parameters are obtained by applying the compensation values in the continuous compensation field to the process parameters according to the adjustment order.

[0041] The main transmission node is determined based on the stress distribution characteristics of the compensation units in the feedback compensation matrix, the compensation transmission link is established from the main transmission node, the compensation values in the compensation transmission link are spatially reconstructed based on the positional association relationship between the compensation units to generate a compensation distribution matrix, including:

[0042] The stress field analysis is performed on the compensation units in the feedback compensation matrix, and the main stress direction and the main stress intensity of each compensation unit are extracted;

[0043] The stress intensity field is constructed according to the distribution of the main stress intensity in the deformation coordinate system, the stress direction field is constructed according to the distribution of the main stress direction in the deformation coordinate system, and the stress distribution characteristics are formed by combining the stress intensity field and the stress direction field;

[0044] The compensation unit with the maximum main stress intensity is selected as the main transmission node based on the stress intensity field, the stress transmission area is established based on the stress direction field, the transmission axis is established in the stress direction of the main transmission node, and the transmission priority is determined based on the distance between the adjacent transmission sub-areas in the stress transmission area and the transmission axis;

[0045] The transmission relationship of the compensation units is determined according to the stress distribution characteristics of the compensation units in the stress transmission area, the transmission relationship is optimized based on the transmission priority, and the optimized transmission relationship is determined as the compensation transmission link;

[0046] The space distance and direction angle between the compensation units are calculated based on the relative positions of the compensation units in the deformation coordinate system, the space distance and direction angle are taken as position correlation parameters, the compensation values of the compensation units in the compensation transmission link are spatially reconstructed according to the position correlation parameters, and a compensation distribution matrix is generated.

[0047] In a second aspect, the present application provides an electronic device, comprising:

[0048] a processor;

[0049] a memory for storing processor-executable instructions;

[0050] The processor is configured to invoke the instructions stored in the memory to execute the method described above.

[0051] In a third aspect, the present application provides a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions are executed by a processor to implement the method described above.

[0052] The present application has the following beneficial effects:

[0053] The present application provides a flexible circuit board three-dimensional deformation prediction and compensation method based on deep learning, which establishes a data set of process parameters and deformation relationship, uses variable density grid units and stress gradient matrix for accurate deformation prediction, and realizes high-precision prediction and compensation of flexible circuit board three-dimensional deformation.

[0054] The present application adopts feedback compensation matrix and compensation transmission link, which can sequentially adjust process parameters according to deformation deviation values, ensures that the influence of stress transmission is considered in the compensation process, effectively avoids new deformation problems caused in the compensation process, and improves the effectiveness and stability of the compensation.

[0055] The present application combines deep learning with material mechanics analysis, and establishes a complete flexible circuit board three-dimensional deformation prediction and compensation system, which not only improves the processing precision of the flexible circuit board, but also reduces the material waste and time cost caused by the traditional trial and error method, and has important practical value for batch production of flexible electronic products. BRIEF DESCRIPTION OF DRAWINGS

[0056] Figure 1 It is a flowchart of the flexible circuit board three-dimensional deformation prediction and compensation method based on deep learning of the present application;

[0057] Figure 2 It is a complete flowchart of the present application for establishing a process parameter-deformation relationship data set based on process parameters and actual three-dimensional deformation data;

[0058] Figure 3 A complete flowchart for constructing a feedback compensation matrix based on a deformation deviation value and adjusting process parameters according to an embodiment of the present application is shown in FIG. 1. DETAILED DESCRIPTION

[0059] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in connection with the accompanying drawings of the embodiments of the present application. Obviously, the described embodiments are only a part but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0060] The technical solutions of the present application will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes can not be described in some embodiments.

[0061] Figure 1 A flowchart of a flexible circuit board three-dimensional deformation prediction and compensation method based on deep learning according to an embodiment of the present application is shown in FIG. 1, which includes the following steps. Figure 1

[0062] Obtaining process parameters in a flexible circuit board production process, and establishing a process parameter-deformation relationship data set according to the process parameters and actual three-dimensional deformation data of the flexible circuit board;

[0063] Constructing a variable-density grid unit along a solder joint distribution direction of the flexible circuit board, calculating a stress gradient matrix based on stress deformation characteristic data of the variable-density grid unit, and constructing a deformation prediction parameter matrix based on the stress gradient matrix;

[0064] Inputting the process parameter-deformation relationship data set into the deformation prediction parameter matrix, calculating three-dimensional deformation prediction data, comparing the actual three-dimensional deformation data with the three-dimensional deformation prediction data, and calculating a deformation deviation value;

[0065] Constructing a feedback compensation matrix based on the deformation deviation value, each compensation unit in the feedback compensation matrix forming a compensation transmission link according to a stress transmission direction, and sequentially adjusting process parameters according to transmission characteristics of the compensation transmission link to obtain adjusted process parameters;

[0066] According to the adjusted process parameters, performing soldering processing on the flexible circuit board to compensate for three-dimensional deformation of the flexible circuit board.

[0067] In an optional embodiment, obtaining process parameters in a flexible circuit board production process, and establishing a process parameter-deformation relationship data set according to the process parameters and actual three-dimensional deformation data of the flexible circuit board includes:​

[0068] Collecting process parameters of the flexible circuit board during the welding process, arranging the process parameters according to the execution time sequence of the welding process according to the collection time of the process parameters, and generating a process parameter time sequence;

[0069] Real-time collection of actual three-dimensional deformation data of the flexible circuit board under the action of the process parameter time sequence, and calculation of the deformation gradient of the actual three-dimensional deformation data at adjacent collection time;

[0070] According to the deformation gradient, the deformation compensation direction is determined, the process parameter difference value of the process parameter time sequence at adjacent time sequence nodes is constructed into a process parameter change matrix, and the deformation compensation coefficient is calculated based on the weight distribution of the temperature process parameter difference value and the pressure process parameter difference value in the process parameter change matrix;

[0071] The deformation compensation direction is applied to the deformation compensation coefficient to generate the deformation compensation amount of each time sequence node;

[0072] The deformation compensation amount is constructed into a compensation adjustment curve in the temperature process parameter associated area and the pressure process parameter associated area, respectively, the process parameter time sequence is compensated in a gradual manner according to the compensation adjustment curve, and the process parameter-deformation relationship data set is established.

[0073] As shown in Figure 2 , the method comprises:

[0074] Collecting process parameters of the flexible circuit board during the welding process, process parameters mainly including welding temperature, welding pressure, welding time, and key parameters such as solder joint position. In the actual production environment, these parameters can be collected in real time through temperature sensors, pressure sensors, time controllers and other devices. The collection accuracy of the welding temperature is usually ±1℃, and the collection frequency is 10Hz; the collection accuracy of the welding pressure is ±0.01MPa, and the collection frequency is 20Hz; the control accuracy of the welding time is ±0.01s. For the welding process of a certain type of flexible circuit board, the typical temperature parameter range is 220℃-260℃, the pressure parameter range is 0.3MPa-0.6MPa, and the time parameter range is 1.5s-3.0s.

[0075] The welding point position is collected by a vision system, and the precision can reach ±0.02 mm. According to the collection time of the process parameters, the process parameter time sequence is arranged according to the execution time sequence of the welding process. For example, in a certain welding process, the temperature is 25°C and the pressure is 0 MPa at t=0 s; the temperature rises to 150°C and the pressure increases to 0.2 MPa at t=1.0 s; the temperature rises to 220°C and the pressure increases to 0.4 MPa at t=2.0 s; the temperature reaches 240°C and the pressure stabilizes at 0.5 MPa at t=3.0 s; the temperature remains at 240°C and the pressure remains at 0.5 MPa at t=5.0 s; the temperature starts to drop and the temperature drops to 220°C and the pressure decreases to 0.3 MPa at t=5.5 s; the temperature drops to 180°C and the pressure decreases to 0.1 MPa at t=6.0 s; the temperature returns to room temperature and the pressure drops to 0 at t=7.0 s. These data are arranged in time sequence to form a complete process parameter time sequence.

[0076] The actual three-dimensional deformation data of the flexible circuit board under the action of the process parameter time sequence is collected in real time. The actual three-dimensional deformation data is collected by a high-precision three-dimensional scanning system, which usually consists of a structured light projector and a high-resolution camera, and can capture the small changes on the surface of the flexible circuit board. The scanning precision can reach ±0.01 mm, and the scanning frequency is 5 Hz, covering the entire surface of the flexible circuit board. During the welding process, a three-dimensional scan is performed every 0.2 s to record the three-dimensional coordinate data of the surface of the flexible circuit board. For example, for a flexible circuit board with a size of 100 mm x 80 mm, about 10,000 three-dimensional coordinate points are obtained at each scanning time to form the three-dimensional deformation data at that time. In different stages of the welding process, the flexible circuit board shows different deformation characteristics. In the heating stage, due to thermal expansion effect, the board surface bends upward, and the maximum deformation can reach 0.5 mm; in the constant temperature stage, due to material softening and pressure effect, the board surface gradually flattens, and the deformation decreases to 0.3 mm; in the cooling stage, due to material shrinkage and release of residual stress, the board surface shows local deformation, and the deformation is unevenly distributed, ranging from 0.1 mm to 0.4 mm.

[0077] The deformation gradient of the actual three-dimensional deformation data at adjacent acquisition time is calculated, and the deformation gradient reflects the rate of change of the deformation variable with time, which is an important parameter for analyzing the dynamic characteristics of deformation. For each acquisition point, the deformation difference between the adjacent two acquisition time is calculated, and then divided by the time interval to obtain the deformation gradient of the point. For example, for the point located at the coordinate (50mm, 40mm), the height at t=2.0s is 0.15mm, and the height at t=2.2s is 0.18mm, then the deformation gradient of the point in this time period is (0.18mm-0.15mm) / 0.2s=0.15mm / s. By calculating the deformation gradient of all acquisition points, the deformation gradient distribution of the entire flexible circuit board is obtained. In different stages of the welding process, the deformation gradient distribution has obvious differences. In the initial heating stage, the deformation gradient is large, which can reach 0.2mm / s-0.3mm / s; in the constant temperature stage, the deformation gradient is small, which is about 0.05mm / s-0.1mm / s; in the cooling stage, the direction of the deformation gradient changes, and the range is between-0.15mm / s and 0.05mm / s.

[0078] The deformation compensation direction is determined according to the deformation gradient, and the deformation compensation direction should be opposite to the direction of the deformation gradient to offset the deformation trend. For each acquisition point, the compensation direction is determined according to the direction of the deformation gradient of the point. For example, if the deformation gradient of a point is positive, it means that the height of the point is increasing, and the compensation direction is negative; on the contrary, if the deformation gradient is negative, it means that the height of the point is decreasing, and the compensation direction is positive. In practical applications, considering the overall deformation characteristics of the flexible circuit board, regions with similar deformation gradients are usually divided into the same compensation region, and the same compensation direction is adopted.

[0079] For example, the region with a deformation gradient between 0.1mm / s and 0.2mm / s is divided into A region, and the compensation direction is negative; the region with a deformation gradient between-0.05mm / s and 0.05mm / s is divided into B region, and the compensation direction is determined according to the trend; the region with a deformation gradient between-0.2mm / s and-0.1mm / s is divided into C region, and the compensation direction is positive.

[0080] The process parameter difference value between adjacent time sequence nodes in the process parameter time sequence is constructed to form a process parameter change matrix, which records the change amount of each process parameter at adjacent time sequence nodes and reflects the dynamic characteristics of the process parameters. For example, from t=2.0s to t=3.0s, the temperature increases from 220℃ to 240℃, and the change amount is 20℃; the pressure increases from 0.4MPa to 0.5MPa, and the change amount is 0.1MPa. These change amounts are organized according to the process parameter type and the time sequence node to form the process parameter change matrix. The rows of the matrix represent different process parameters, and the columns represent different time sequence node pairs. By analyzing the process parameter change matrix, the process parameters and time sequence stages that have a greater impact on deformation can be identified.

[0081] The deformation compensation coefficient is calculated based on the weight distribution of the temperature process parameter difference and the pressure process parameter difference in the process parameter variation matrix. The deformation compensation coefficient is a parameter that describes the degree of influence of process parameter variation on deformation, and is used to quantify the size of the compensation adjustment. By analyzing historical data and experimental results, the weight of the temperature parameter and the pressure parameter on the deformation is determined. In this embodiment, the weight of the temperature parameter is 0.7, and the weight of the pressure parameter is 0.3, indicating that the influence of temperature change on deformation is 2.33 times that of pressure change. According to these weights, combined with the differences in the process parameter variation matrix, the deformation compensation coefficient is calculated. For example, for a certain timing node, the temperature change is 20℃, and the pressure change is 0.1MPa, then the deformation compensation coefficient is 20℃×0.7+0.1MPa×0.3=14.03 (after unit conversion). The size of the deformation compensation coefficient reflects the degree of compensation adjustment required, and the larger the coefficient, the greater the compensation adjustment required.

[0082] The deformation compensation direction is applied to the deformation compensation coefficient to generate the deformation compensation amount for each timing node. The deformation compensation amount is the actual adjustment amount applied to the process parameter, which is determined by the deformation compensation direction and the deformation compensation coefficient. For each timing node, the deformation compensation direction of the node is multiplied by the deformation compensation coefficient to obtain the deformation compensation amount. For example, if the deformation compensation direction of a certain timing node is negative and the deformation compensation coefficient is 14.03, then the deformation compensation amount is -14.03. According to the size and direction of the deformation compensation amount, the process parameters are adjusted accordingly to offset or reduce the deformation. In practical applications, the deformation compensation amount is usually converted into specific process parameter adjustment values, such as temperature adjustment values and pressure adjustment values. For example, a deformation compensation amount of -14.03 corresponds to a temperature decrease of 10℃ and a pressure decrease of 0.05MPa adjustment.

[0083] The deformation compensation amount is used to construct a compensation adjustment curve in the temperature process parameter associated area and the pressure process parameter associated area respectively. The compensation adjustment curve describes how the deformation compensation amount is converted into specific process parameter adjustment values, and how these adjustment values change over time. The temperature process parameter associated area refers to the area where deformation is mainly affected by temperature, such as the high temperature area around the solder joints; the pressure process parameter associated area refers to the area where deformation is mainly affected by pressure, such as the stress concentration area around the force points. For each associated area, a corresponding compensation adjustment curve is constructed according to the deformation characteristics and process parameter sensitivity of the area. For example, for the temperature process parameter associated area, the compensation adjustment curve shows a linear relationship between the deformation compensation amount and the temperature adjustment value, with a slope of 0.5℃ / unit compensation amount; for the pressure process parameter associated area, the compensation adjustment curve shows a nonlinear relationship between the deformation compensation amount and the pressure adjustment value, with a smaller slope in the small compensation amount range and a larger slope in the large compensation amount range.

[0084] According to the compensation adjustment curve, the process parameter time sequence is regionally and progressively compensated, and a process parameter-deformation relationship dataset is established. The regional and progressive compensation refers to adopting different compensation strategies according to the deformation characteristics and compensation requirements of different regions, and gradually adjusting the process parameters to achieve accurate compensation. For each time sequence node, according to the deformation compensation amount of the node, the corresponding compensation adjustment curve is queried to determine the temperature adjustment value and the pressure adjustment value. Then, the adjustment values are applied to the original process parameters to obtain the compensated process parameters. For example, for the time sequence node of t=3.0s, the original temperature is 240℃, and the deformation compensation amount is-14.03. The temperature adjustment value obtained by querying the temperature compensation adjustment curve is-7℃, and the compensated temperature is 233℃. Through similar calculation, the compensated process parameters of all time sequence nodes are obtained to form a complete process parameter-deformation relationship dataset. The dataset records the deformation conditions of the flexible circuit board under different process parameter combinations, including original process parameters, compensated process parameters, deformation amount, deformation gradient and other information. For example, a data item contains the following information: time t=3.0s, original temperature 240℃, original pressure 0.5MPa, compensated temperature 233℃, compensated pressure 0.48MPa, deformation amount 0.3mm, and deformation gradient 0.15mm / s.

[0085] By analyzing the rules and trends in the dataset, a mapping relationship between the process parameters and the deformation can be established to provide data support for deformation prediction. At the same time, the compensation experience in the dataset can also be used to guide the design of process parameters for new products, improving production efficiency and product quality. In actual application, with the continuous accumulation of production data, the process parameter-deformation relationship dataset will be continuously expanded and optimized, and the accuracy of deformation prediction and compensation will also be continuously improved.

[0086] The method provided by the application realizes accurate process parameter compensation and deformation control by collecting process parameters and deformation data in real time, analyzing deformation gradient and process parameter changes, calculating deformation compensation coefficients and compensation amounts, and constructing compensation adjustment curves. Compared with the traditional method, the method considers the dynamic characteristics of deformation and the comprehensive influence of process parameters in the welding process, and the compensation effect is more accurate, which can meet the strict requirements of high-precision electronic products on deformation control of flexible circuit boards.

[0087] In an optional implementation, a variable-density grid unit is constructed along the distribution direction of the solder joints of the flexible circuit board, a stress gradient matrix is calculated based on stress deformation characteristic data of the variable-density grid unit, and a deformation prediction parameter matrix is constructed based on the stress gradient matrix, including:

[0088] generate variable-density grid cells along the solder distribution direction of the flexible circuit board, the variable-density grid cells being set as micro-scale grid cells in the solder distribution area and set as macro-scale grid cells in the non-solder area;

[0089] acquire first stress deformation characteristic data in the micro-scale grid cells, the first stress deformation characteristic data representing stress distribution and deformation trend in the solder area, and acquire second stress deformation characteristic data in the macro-scale grid cells, the second stress deformation characteristic data representing stress distribution and deformation trend in the non-solder area;

[0090] perform multi-scale deformation field reconstruction based on the grid density distribution of the first stress deformation characteristic data and the second stress deformation characteristic data to obtain a stress field reconstruction matrix;

[0091] calculate stress gradients of the variable-density grid cells based on the stress field reconstruction matrix, and generate a stress gradient matrix according to the distribution positions of the variable-density grid cells;

[0092] perform feature decoupling on the stress field reconstruction matrix and the stress gradient matrix to extract stress deformation correlation features, and construct a deformation prediction parameter matrix according to the stress deformation correlation features.

[0093] generate variable-density grid cells along the solder distribution direction of the flexible circuit board, the variable-density grid cells being grid cells with different densities in different areas, wherein micro-scale grid cells are set in the solder distribution area, and macro-scale grid cells are set in the non-solder area. The edge length of the micro-scale grid cells is usually 0.05 mm to 0.1 mm, which is used to accurately describe the stress concentration and deformation details near the solder; the edge length of the macro-scale grid cells is usually 0.5 mm to 2 mm, which is used to describe the overall stress distribution and deformation trend in the non-solder area.

[0094] In actual application, for a flexible circuit board with a size of 100 mm x 80 mm, the number of micro-scale grid cells in the solder area (with a diameter of about 5 mm) is about 10,000, and the number of macro-scale grid cells in the non-solder area is about 5,000. In the generation process of the variable-density grid cells, the distribution characteristics of the solder need to be considered. For common electronic components such as chips, resistors, and capacitors, the solder is usually arranged in a certain direction. For example, for a certain type of chip, the solder is arranged in a rectangle along the edge with a solder spacing of 0.3 mm; for a resistor, the solder is arranged linearly with a solder spacing of 0.5 mm. According to these solder distribution characteristics, the grid cells with gradually changing density are generated along the solder arrangement direction, and the grid density is highest at the solder, and gradually decreases with the increasing distance from the solder.

[0095] The first stress deformation characteristic data of the micro-scale grid unit includes the principal stress size, the principal stress direction, the strain and the deformation, etc. The acquisition method can adopt finite element analysis or experimental measurement. In the finite element analysis, based on the material properties, the geometric dimensions and the boundary conditions of the flexible circuit board, the mechanical analysis is performed on the micro-scale grid unit, and the stress and deformation of each unit are calculated.

[0096] In the actual case, for the flexible circuit board after welding, the principal stress size of the micro-scale grid unit near the welding point is usually between 30 MPa and 50 MPa, the principal stress direction forms an angle of 30° to 45° with the direction of the welding point connecting line, and the deformation is between 0.05 mm and 0.2 mm. For example, for the micro-scale grid unit near the chip welding point, the principal stress size is 42 MPa, the principal stress direction is 35°, and the deformation is 0.12 mm. These data reflect the stress concentration phenomenon and the local deformation characteristics of the welding point area.

[0097] The second stress deformation characteristic data of the macro-scale grid unit also includes the principal stress size, the principal stress direction, the strain and the deformation, etc., but the spatial resolution is lower, mainly reflecting the overall deformation trend. In the non-welding point area, the stress distribution is relatively uniform, the principal stress size is usually between 5 MPa and 15 MPa, and the deformation is between 0.01 mm and 0.05 mm. For example, the macro-scale grid unit 10 mm away from the welding point has a principal stress size of 8 MPa, a principal stress direction of 60°, and a deformation of 0.03 mm. The second stress deformation characteristic data is of great significance for understanding the overall deformation behavior of the flexible circuit board, and can complement the first stress deformation characteristic data to comprehensively describe the stress distribution and deformation trend of the flexible circuit board.

[0098] Based on the grid density distribution of the first stress deformation characteristic data and the second stress deformation characteristic data, the multi-scale deformation field reconstruction is performed to obtain the stress field reconstruction matrix. The multi-scale deformation field reconstruction is the process of fusing stress deformation characteristic data of different scales to generate a continuous stress field covering the entire flexible circuit board. In this embodiment, the multi-scale deformation field reconstruction adopts a density weighted fusion method, which performs weighted average on the stress deformation characteristic data according to the density distribution of the grid unit. Specifically, for any point P in the reconstruction area, the distance from the point P to the surrounding micro-scale grid unit and macro-scale grid unit is calculated, the weight is determined based on the distance, and then the weighted average calculation is performed.

[0099] For example, point P has 3 micro-scale grid units M1, M2, M3 around it, with distances of 0.2 mm, 0.3 mm, 0.5 mm, and principal stresses of 40 MPa, 38 MPa, 35 MPa, respectively; and 2 macro-scale grid units L1, L2, with distances of 1.0 mm, 1.5 mm, and principal stresses of 10 MPa, 8 MPa, respectively. Using the inverse distance weighting method, the weight of M1 is 5, M2 is 3.33, M3 is 2, L1 is 1, and L2 is 0.67, and the total weight is 12. The reconstructed principal stress of point P is (40x5+38x3.33+35x2+10x1+8x0.67) / 12=34.4 MPa.

[0100] Through similar calculations, the stress distribution of all points on the entire flexible circuit board is obtained to form a stress field reconstruction matrix. The size of the stress field reconstruction matrix is m x n x k, where m and n correspond to the number of discrete points in the length and width directions of the flexible circuit board, respectively, and k corresponds to the number of stress parameters stored in the matrix. In this embodiment, m=200, n=160, and k=6, corresponding to a spatial resolution of 0.5 mm x 0.5 mm, and the stored parameters include the principal stress magnitude and principal stress direction in three directions.

[0101] Based on the stress field reconstruction matrix, the stress gradient of the variable-density grid unit is calculated, and the stress gradient matrix is generated according to the distribution position of the variable-density grid unit. The stress gradient reflects the rate of change of stress in space and is an important parameter for predicting deformation. The stress gradient is calculated using the central difference method, which differentiates the stress data in the stress field reconstruction matrix. In the x direction, the stress gradient is equal to the stress difference between the two adjacent points divided by the distance between the two points; the calculation method for the y and z directions is similar. For example, for the point at position (100, 80, 3) in the stress field reconstruction matrix, the x-direction stress gradient is calculated as [stress(101,80,3)-stress(99,80,3)] / (2x0.5mm)=10MPa / mm.

[0102] For boundary points, forward or backward difference is used for calculation, and by calculating the stress gradient of all points, the stress gradient matrix is obtained. The stress gradient matrix has the same spatial structure as the stress field reconstruction matrix, but each position stores the stress gradient value instead of the stress value. In practical applications, the stress gradient near the solder joint is usually large, which can reach 20 MPa / mm to 50 MPa / mm; the stress gradient in the area far from the solder joint is small, which is usually 1 MPa / mm to 5 MPa / mm. The stress gradient matrix directly reflects the non-uniformity of the stress distribution, which helps to identify areas with high deformation risk.

[0103] The stress field reconstruction matrix and the stress gradient matrix are decoupled to extract stress deformation correlation features. Decoupling refers to decomposing the complex stress deformation relationship into multiple independent feature components to more accurately describe the correlation between stress and deformation. In this embodiment, principal component analysis is used for decoupling. The stress field reconstruction matrix and the stress gradient matrix are combined to form a feature matrix, and principal component analysis is performed on the feature matrix to extract the main feature components. Usually, the first 5 to 10 principal components are selected, which can explain more than 80% of the total variance.

[0104] For example, for the feature matrix of a certain flexible circuit board, the first 5 principal components explain 40%, 25%, 10%, 8% and 5% of the total variance, respectively, which cumulatively explains 88% of the variance. The feature vectors corresponding to these 5 principal components reflect the stress concentration features of the solder area, the bending deformation features of the board edge, the thermal deformation features caused by temperature gradient, the local deformation features caused by material inhomogeneity, and the dynamic deformation features caused by vibration, respectively. By analyzing the physical meaning of these feature vectors, a correlation model between stress and deformation is established, and stress deformation correlation features are extracted.

[0105] Stress deformation correlation features are quantitative parameters that describe how stress affects deformation, including stress sensitivity coefficients, stress direction factors, gradient influence coefficients, etc. For example, for the solder area, the stress sensitivity coefficient is 0.005 mm / MPa, indicating that for every 1 MPa increase in stress, the deformation increases by 0.005 mm; the stress direction factor is 0.8, indicating that the consistency of the principal stress direction and the deformation direction is 80%; and the gradient influence coefficient is 0.002 mm² / MPa, indicating that for every 1 MPa / mm increase in stress gradient, the deformation increases by 0.002 mm.

[0106] A deformation prediction parameter matrix is constructed according to the stress deformation correlation features. The deformation prediction parameter matrix is a data structure used to store various parameters required for predicting deformation, including stress sensitivity coefficients, stress direction factors, gradient influence coefficients, etc. The spatial structure of the deformation prediction parameter matrix is the same as that of the stress field reconstruction matrix, but each position stores a prediction parameter instead of a stress value. When constructing the deformation prediction parameter matrix, according to the stress deformation correlation features of different regions, the corresponding prediction parameters are set.

[0107] For the high stress area near the welding point, the stress sensitivity coefficient is higher, usually between 0.004mm / MPa and 0.008mm / MPa; for the non-welding point area, the stress sensitivity coefficient is lower, usually between 0.001mm / MPa and 0.003mm / MPa. At the same time, considering the influence of stress direction and stress gradient, the prediction parameters are determined comprehensively. The deformation prediction parameter matrix also contains the information of the stress transmission channel, which is used to describe the path and mode of stress transmission in the flexible circuit board. The stress transmission channel usually extends along the high stiffness direction of the material and forms a connection between the welding points. For example, for the area between two adjacent welding points, the direction of the stress transmission channel is consistent with the direction of the line connecting the two welding points, and the transmission efficiency is 85%, indicating that the stress attenuates by 15% during transmission.

[0108] After the construction of the deformation prediction parameter matrix is completed, it can be used for subsequent deformation prediction calculation. The process parameter-deformation relationship data set is input into the deformation prediction parameter matrix, and the three-dimensional deformation prediction data is calculated through the prediction model in the parameter matrix. In actual application, the deformation prediction parameter matrix can accurately predict the deformation of the flexible circuit board under different process conditions, and the prediction accuracy reaches more than 90%, which provides strong support for the design optimization and process parameter adjustment of the flexible circuit board.

[0109] The method of the present application realizes accurate prediction of the three-dimensional deformation of the flexible circuit board by constructing variable-density grid elements, obtaining multi-scale stress deformation feature data, performing multi-scale deformation field reconstruction, calculating stress gradient matrix and constructing deformation prediction parameter matrix, etc., which provides technical support for deformation compensation and reliability improvement of the flexible circuit board.

[0110] In an alternative embodiment, multi-scale deformation field reconstruction is performed based on the grid density distribution of the first stress deformation feature data and the second stress deformation feature data to obtain a stress field reconstruction matrix, comprising:

[0111] The microscale grid elements and the macroscale grid elements are divided into deformation reconstruction regions according to the density gradient size, the density change rate of the deformation reconstruction regions is calculated, and a reconstruction feature vector is constructed according to the density change rate;

[0112] A spiral reconstruction path is generated based on the density distribution of the reconstruction feature vector, the geometric parameters of the spatial configuration of the spiral reconstruction path match the density change process of the reconstruction feature vector, and the first stress deformation feature data and the second stress deformation feature data are density-reconstructed and mapped along the spiral reconstruction path to generate reconstruction-enhanced data;

[0113] A reconstruction data transmission channel is established inside the morphing reconstruction region, the direction of the reconstruction data transmission channel is consistent with the density gradient direction of the reconstruction feature vector, and the reconstruction enhancement data is transmitted along the reconstruction data transmission channel to obtain transmitted reconstruction enhancement data;

[0114] The transmitted reconstruction enhancement data is used to generate a stress field reconstruction matrix according to the density gradient direction of the reconstruction feature vector.

[0115] The micro-scale grid cells and the macro-scale grid cells are divided into morphing reconstruction regions according to the density gradient size, the micro-scale grid cells are usually small in size and high in density, and are suitable for describing local details; the macro-scale grid cells are large in size and low in density, and are suitable for describing overall trends. In this embodiment, the size of the micro-scale grid cells is 0.1 mm x 0.1 mm, and the size of the macro-scale grid cells is 2 mm x 2 mm. For a flexible circuit board with a size of 100 mm x 80 mm, the total number of micro-scale grid cells is about 800,000, and the total number of macro-scale grid cells is about 2,000. According to the density gradient size of the grid cells, the region with a density gradient greater than 0.5 cells / mm 2 is divided into a high-gradient morphing reconstruction region, the region with a density gradient between 0.2 and 0.5 cells / mm 2 is divided into a medium-gradient morphing reconstruction region, and the region with a density gradient less than 0.2 cells / mm 2 is divided into a low-gradient morphing reconstruction region.

[0116] The density change rate of the morphing reconstruction region is calculated to construct a reconstruction feature vector, and the density change rate describes the change of the grid density along a specific direction, which is a key parameter for constructing the reconstruction feature vector. In this embodiment, the density change rate is calculated along the x-axis, y-axis and z-axis directions. For the high-gradient morphing reconstruction region, the average density change rate along the x-axis direction is 0.8 cells / mm 2 , the average density change rate along the y-axis direction is 0.6 cells / mm 2 , and the average density change rate along the z-axis direction is 0.4 cells / mm 2 ; for the medium-gradient morphing reconstruction region, the density change rates along the three directions are 0.4, 0.3 and 0.2 cells / mm 2 ; and for the low-gradient morphing reconstruction region, the density change rates along the three directions are 0.15, 0.12 and 0.08 cells / mm 2 / mm. Based on these density change rates, a reconstruction feature vector is constructed. The reconstruction feature vector is a three-dimensional vector, whose direction points to the direction of the fastest density increase, and whose size equals the density change rate in that direction. For example, at a certain point in the high-gradient deformation reconstruction region, if the density change rate in the x-axis direction is 0.9 units / mm 2 , the density change rate in the y-axis direction is 0.5 units / mm 2 , and the density change rate in the z-axis direction is 0.3 units / mm 2 , then the reconstruction feature vector of the point has an angle of 29.1 degrees with the x-axis, an angle of 62.5 degrees with the y-axis, and an angle of 73.2 degrees with the z-axis, and the vector size is 1.07 units / mm 2 .

[0117] Based on the density distribution of the reconstruction feature vector, a spiral reconstruction path is generated, which is a special spatial curve whose geometric parameters (such as pitch, radius, rotation angular velocity, etc.) match the density change process of the reconstruction feature vector. In this embodiment, the basic form of the spiral reconstruction path is a three-dimensional spiral line, the central axis of the spiral is consistent with the average direction of the reconstruction feature vector, the pitch of the spiral is inversely proportional to the density change rate, and the radius of the spiral is proportional to the local grid density. For example, in the high-gradient deformation reconstruction region, the average direction of the reconstruction feature vector is (0.7, 0.5, 0.3), the average density change rate is 0.8 units / mm 2 , and the average grid density is 25 units / mm 2 , then the central axis direction of the spiral reconstruction path of the region is (0.7, 0.5, 0.3), the pitch is 1 / 0.8 = 1.25 mm, and the radius is 25 x 0.02 = 0.5 mm. In the medium-gradient and low-gradient deformation reconstruction regions, similar methods are used to generate spiral reconstruction paths, forming a continuous path network covering the entire deformation reconstruction region.

[0118] The first stress deformation feature data and the second stress deformation feature data are density-reconstructed and mapped along a spiral reconstruction path to generate reconstruction-enhanced data. The first stress deformation feature data is from micro-scale grid cells and contains local detailed information; the second stress deformation feature data is from macro-scale grid cells and contains overall trend information. Density-reconstruction mapping is a data fusion technique that maps data of different scales into a unified space according to specific rules. In this embodiment, a weighted average method is used for density-reconstruction mapping. For each point on the spiral reconstruction path, the distances to the surrounding micro-scale grid cells and macro-scale grid cells are calculated, the weights are determined based on the distances, and then the weighted average value is calculated as the reconstruction-enhanced data of the point. The weight calculation uses the inverse distance weighting method, that is, the closer the distance, the greater the weight, and the farther the distance, the smaller the weight. For example, for a point P on the spiral reconstruction path, there are three micro-scale grid cells M1, M2, M3 around it, with distances of 0.2 mm, 0.3 mm, and 0.5 mm, and stress deformation feature data of 35 MPa, 32 MPa, and 30 MPa, respectively; there are two macro-scale grid cells L1 and L2, with distances of 1.0 mm and 1.5 mm, and stress deformation feature data of 28 MPa and 26 MPa, respectively. The weights of each cell are calculated: M1 is 1 / 0.2 = 5, M2 is 1 / 0.3 = 3.33, M3 is 1 / 0.5 = 2, L1 is 1 / 1.0 = 1, and L2 is 1 / 1.5 = 0.67, and the total weight is 5 + 3.33 + 2 + 1 + 0.67 = 12. The reconstruction-enhanced data of point P is (35 × 5 + 32 × 3.33 + 30 × 2 + 28 × 1 + 26 × 0.67) / 12 = 32.4 MPa. Through similar calculations, the reconstruction-enhanced data of all points on the spiral reconstruction path is obtained.

[0119] A reconstruction data transmission channel is established inside the deformation reconstruction region. The reconstruction data transmission channel is a path for data transmission in space, and its direction is consistent with the density gradient direction of the reconstruction feature vector, and is used to transmit reconstruction-enhanced data. In this embodiment, the reconstruction data transmission channel adopts a tree structure and extends from a high-density area to a low-density area, with branch points located at positions where the density gradient direction changes significantly. For a high-gradient deformation reconstruction region, the main trunk of the transmission channel extends along the average density gradient direction, with a length of about 20 mm, and a branch point is set every 2 mm, and each branch point extends 3-5 branches with a branch length of 2-5 mm; for medium-gradient and low-gradient deformation reconstruction regions, the structural parameters of the transmission channel are correspondingly reduced. For example, in a certain high-gradient region, the average density gradient direction is (0.6, 0.4, 0.3), the main trunk length is 18 mm, the branch points are located at positions 2 mm, 4 mm, 6 mm, etc. from the starting point of the main trunk, each branch point extends 4 branches, and the average branch length is 3.5 mm.

[0120] The reconstruction enhancement data is transmitted along the reconstruction data transmission channel to obtain transmitted reconstruction enhancement data. Data transmission refers to the process of transmitting reconstruction enhancement data on the spiral reconstruction path to the entire morphing reconstruction region. In this embodiment, the data transmission adopts a decay transmission model, and the transmission strength decays with increasing distance. Starting from the starting point of the transmission channel, the reconstruction enhancement data is transmitted outward along the transmission channel, and the transmission coefficient is 0.9, that is, the data strength decays by 10% for every 1mm of transmission. For example, the reconstruction enhancement data at the starting point of the transmission channel is 40MPa, after 1mm of transmission, it is 40*0.9=36MPa, after 2mm of transmission, it is 36*0.9=32.4MPa, and so on. During the transmission process, if a branch point is encountered, the data is split to each branch, and the total transmission amount remains unchanged. In this way, the reconstruction enhancement data is transmitted to every position in the morphing reconstruction region to obtain the transmitted reconstruction enhancement data.

[0121] The transmitted reconstruction enhancement data is used to generate a stress field reconstruction matrix according to the density gradient direction of the reconstruction feature vector. The stress field reconstruction matrix is a data structure for describing stress distribution, and its arrangement direction is consistent with the density gradient direction of the reconstruction feature vector. In this embodiment, the stress field reconstruction matrix adopts a three-dimensional matrix structure, and the three dimensions of the matrix correspond to the x, y, and z directions of the space, respectively. The value of the matrix element represents the stress value at the corresponding position. The size of the matrix is determined according to the size of the flexible circuit board and the analysis accuracy, and is usually 200*160*20, corresponding to a spatial resolution of 0.5mm*0.5mm*0.05mm. The transmitted reconstruction enhancement data is mapped to the corresponding position in the matrix to form a complete stress field reconstruction matrix. For example, the point located at the coordinate (25mm, 30mm, 0.5mm) has a transmitted reconstruction enhancement data of 35MPa, and this value is stored at the (50, 60, 10) position of the matrix (assuming that the matrix index starts from 0). For positions without direct calculation data, an interpolation method is used to obtain the value. Common interpolation methods include linear interpolation, bilinear interpolation, and trilinear interpolation.

[0122] The stress field reconstruction matrix can be used for subsequent stress analysis and deformation prediction. By analyzing the stress distribution characteristics in the matrix, stress concentration areas and potential deformation risk points can be identified. For example, in the stress field reconstruction matrix of a certain flexible circuit board, it is found that there is a significant stress concentration phenomenon around the main chip, with a maximum stress value of 45MPa, which is much higher than the 20-25MPa in other areas, indicating that this area is a potential deformation risk point. Based on this finding, the chip packaging process can be optimized, or a reinforcing structure can be added in this area to reduce the stress concentration.

[0123] The method fully utilizes the advantages of micro-scale and macro-scale grid units through multi-scale deformation field reconstruction, and realizes accurate description of stress distribution of the flexible circuit board. Compared with the traditional single-scale analysis method, the method significantly improves the analysis accuracy while maintaining the computing efficiency, especially the ability to depict the stress concentration area. In practical applications, the method has been successfully applied to the analysis of various types of flexible circuit boards, and the prediction accuracy has been improved by more than 35%, providing strong support for the design optimization and reliability improvement of the flexible circuit board.

[0124] In an optional embodiment, the process parameter-deformation relationship dataset is input into the deformation prediction parameter matrix, and three-dimensional deformation prediction data is calculated. The actual three-dimensional deformation data of the flexible circuit board is compared with the three-dimensional deformation prediction data, and a deformation deviation value is calculated, including:

[0125] The process parameter-deformation relationship dataset is input into the deformation prediction parameter matrix, and the process parameters are quantitatively processed based on the stress transmission channels in the deformation prediction parameter matrix. The deformation prediction flow field is generated based on the quantized process parameters in the stress transmission channels.

[0126] The three-dimensional deformation prediction data is obtained by expanding the deformation prediction flow field in three-dimensional space, and a three-dimensional deformation prediction surface is generated based on the three-dimensional deformation prediction data.

[0127] The curvature constraint condition is constructed based on the deformation amount ratio between adjacent feature positions in the actual three-dimensional deformation data, and the three-dimensional deformation real surface is generated by tensor completion of the actual three-dimensional deformation data. The curvature constraint condition is used to limit the spatial continuous features of the actual three-dimensional deformation data.

[0128] A continuous transition surface is constructed between the three-dimensional deformation prediction surface and the three-dimensional deformation real surface, and a hyperboloid interference graph is constructed based on the hyperbolic transition features of the continuous transition surface.

[0129] The wave peak position and the wave valley position of the hyperboloid interference graph are extracted, and the spatial coordinate difference between the wave peak position and the wave valley position is calculated as a deformation deviation value.

[0130] The process parameter-deformation relationship dataset is input into the deformation prediction parameter matrix. The process parameter-deformation relationship dataset contains a large amount of historical production data, recording the deformation conditions of the flexible circuit board under different process parameter combinations. In this embodiment, the process parameters mainly include soldering temperature, soldering pressure, soldering time, solder distribution, and material properties, etc. For example, in a certain production, when the soldering temperature is 240°C, the soldering pressure is 0.5 MPa, and the soldering time is 2.5 s, the deformation amount of a certain model of flexible circuit board at a specific position is 0.22 mm. By collecting a large amount of such data, a process parameter-deformation relationship dataset containing 5000 groups of data is established. The deformation prediction parameter matrix is calculated in advance by the stress deformation characteristic data of the variable density grid unit, which contains the information of the stress transmission channel. The stress transmission channel describes the path of stress transmission in the flexible circuit board, which usually extends along the high stiffness direction of the material.

[0131] The process parameters are hierarchically quantized based on the stress transmission channels in the deformation prediction parameter matrix. Hierarchical quantization is the process of converting continuous process parameter values into discrete levels, which facilitates transmission calculation in the stress transmission channel. In this embodiment, the soldering temperature is divided into 5 levels (220-230°C is level 1, 230-240°C is level 2, 240-250°C is level 3, 250-260°C is level 4, and 260-270°C is level 5), the soldering pressure is divided into 4 levels (0.3-0.4 MPa is level 1, 0.4-0.5 MPa is level 2, 0.5-0.6 MPa is level 3, and 0.6-0.7 MPa is level 4), and the soldering time is divided into 3 levels (1.5-2.0 s is level 1, 2.0-2.5 s is level 2, and 2.5-3.0 s is level 3). Through this hierarchical quantization, the process parameters can be effectively transmitted within the stress transmission channel.

[0132] Based on the quantized process parameters, a deformation prediction flow field is generated within the stress transmission channel. The deformation prediction flow field describes the dynamic process of deformation propagation in the flexible circuit board, which can be regarded as the flow field of stress transmission in the material causing deformation. In this embodiment, the stress transmission channel is divided into multiple sub-channels, each corresponding to different transmission characteristics. The process parameters are transmitted through these sub-channels, generating deformation flows of different intensities. The intensity of the deformation flow is related to the level of the process parameters and the transmission coefficient of the stress transmission channel. For example, the process parameter combination of soldering temperature level 3, soldering pressure level 2, and soldering time level 3 generates a deformation flow intensity of 0.25 mm / mm in a sub-channel with a transmission coefficient of 0.85. By calculating the deformation flow intensity of all sub-channels, a complete deformation prediction flow field is generated.

[0133] The three-dimensional deformation prediction data is obtained by expanding the deformation prediction flow field in three-dimensional space. The deformation prediction flow field is a flow field on a two-dimensional plane, which needs to be converted into three-dimensional deformation data in three-dimensional space. The conversion process takes into account factors such as the material properties, thickness distribution and solder location of the flexible circuit board. In this embodiment, the layered integration method is used to convert the flow field into three-dimensional deformation data. The flexible circuit board is divided into 10 layers, each layer is 0.02mm thick, the deformation of each layer is calculated, and then integrated to obtain the overall deformation. For example, a point located at coordinates (25mm, 30mm) has a deformation prediction flow field intensity of 0.25mm / mm, after layered integration calculation, the three-dimensional deformation prediction value of the point is 0.20mm, the direction is 45°, and the height increases by 0.12mm. By calculating the three-dimensional deformation prediction value of all points on the flexible circuit board, complete three-dimensional deformation prediction data is obtained.

[0134] The three-dimensional deformation prediction surface is generated based on the three-dimensional deformation prediction data. The three-dimensional deformation prediction surface is a continuous surface that describes the predicted deformation state of the flexible circuit board, and is generated by interpolating the discrete three-dimensional deformation prediction data. In this embodiment, the radial basis function interpolation method is used to generate the three-dimensional deformation prediction surface. The deformation feature points are selected as the interpolation base points, the interpolation radius is 5mm, and the smoothing coefficient is 0.8. In this way, a three-dimensional deformation prediction surface covering the entire flexible circuit board is generated, with a resolution of 0.5mm x 0.5mm. The prediction surface intuitively shows the expected deformation state of the flexible circuit board under given process parameters.

[0135] The curvature constraint condition is constructed based on the deformation ratio between adjacent feature positions in the actual three-dimensional deformation data. The actual three-dimensional deformation data is measured by a high-precision three-dimensional scanner, and records the actual deformation of the feature points on the flexible circuit board. Due to measurement limitations, the actual deformation data is usually discrete and needs to be completed by the curvature constraint condition. The curvature constraint condition limits the spatial continuity of the deformation surface, ensuring the smoothness of the deformation surface. In this embodiment, the deformation ratio between adjacent feature points is calculated as the curvature constraint condition. For example, the deformation of feature point A is 0.18mm, the deformation of feature point B is 0.15mm, and the distance between the two points is 3mm, so the deformation ratio is 0.15 / 0.18=0.833, and the curvature constraint condition is set to vary within the range of 0.8-0.9.

[0136] The actual three-dimensional deformation data is tensor completed to generate a three-dimensional deformation measured surface. Tensor completion is a data completion technique used to infer the values of unknown data points based on known data points. In this embodiment, the actual three-dimensional deformation data is organized as a three-dimensional tensor, and the missing data points are completed by a tensor completion algorithm. The completion process takes into account the curvature constraint condition, ensuring the continuity and reasonableness of the completion result. For example, given feature point A (25mm, 30mm, 0.18mm) and feature point B (28mm, 30mm, 0.15mm), the deformation of intermediate point C (26.5mm, 30mm,?) needs to be completed. According to the curvature constraint condition, the deformation of point C should be 0.166mm. By similar calculation, all missing data points are completed to generate a complete three-dimensional deformation measured surface.

[0137] A continuous transition surface is constructed between the three-dimensional deformation prediction surface and the three-dimensional deformation measured surface, which describes the transition process from the prediction surface to the measured surface, and is used to analyze the differences between the two. The construction process takes into account the spatial position and shape characteristics of the two surfaces, ensuring the smoothness of the transition. In this embodiment, a weighted average method is used to construct the continuous transition surface. For each point P on the surface, its coordinates on the transition surface are the weighted average of the prediction surface coordinates and the measured surface coordinates, with the weight continuously changing from 0 to 1. For example, when the weight is 0.3, the point coordinates on the transition surface are 70% of the prediction surface coordinates plus 30% of the measured surface coordinates. By changing the weight, a series of transition surfaces are generated, forming a continuous transition process.

[0138] Based on the hyperbolic transition characteristics of the continuous transition surface, a hyperboloid interference map is constructed. The hyperboloid interference map is a visualization technique used to visually display the differences between two surfaces. The hyperbolic transition characteristics refer to the hyperbolic characteristics exhibited by the transition surface in certain directions, which are manifested as interference fringes in the interference map. In this embodiment, the interference threshold is set to 0.02mm, i.e. when the height difference between two adjacent points on the transition surface reaches 0.02mm, an interference fringe is formed. By calculating the height difference of all points on the transition surface, a complete hyperboloid interference map is generated. The fringe density in the interference map reflects the size of the deformation difference, and the fringe direction reflects the direction of the deformation difference.

[0139] The peak positions and the valley positions of the hyperboloid interference figure are extracted. The peak positions correspond to the bright fringes in the interference figure, representing the areas where the deformation prediction value is greater than the measured value. The valley positions correspond to the dark fringes in the interference figure, representing the areas where the deformation prediction value is less than the measured value. The extraction process uses image processing techniques to identify the bright and dark fringes in the interference figure and determine their spatial positions. In this embodiment, the interference figure resolution is 0.2 mm, and the identification thresholds are 80% and 20% of the brightness value, corresponding to the peak and valley positions, respectively. In this way, all the peak and valley positions in the interference figure are extracted, containing about 300 peak points and 280 valley points.

[0140] The spatial coordinate difference between the peak positions and the valley positions is calculated as the deformation deviation value, which directly reflects the deviation between the deformation prediction and the actual measurement. The calculation process takes into account the distance and direction factors in three-dimensional space, providing a comprehensive assessment of the spatial distribution of deformation deviation. In this embodiment, for each pair of adjacent peak points and valley points, the coordinate difference in x, y, and z directions, as well as the Euclidean distance, is calculated. For example, the coordinate difference between peak point P1 (25.2 mm, 30.1 mm, 0.22 mm) and valley point P2 (25.4 mm, 30.3 mm, 0.18 mm) is (0.2 mm, 0.2 mm, -0.04 mm), and the Euclidean distance is 0.28 mm. By statistically analyzing the coordinate differences of all peak-valley pairs, the spatial distribution characteristics of the deformation deviation are obtained.

[0141] The calculation results of the deformation deviation value are used for subsequent feedback compensation process. In this embodiment, the maximum deformation deviation value is 0.32 mm, occurring near the main chip solder joint; the minimum deformation deviation value is 0.05 mm, occurring in the edge area of the board; and the average deformation deviation value is 0.18 mm. These deviation values directly reflect the accuracy of the deformation prediction and provide an important basis for deformation compensation. By analyzing the spatial distribution of the deformation deviation value, the weak links in the deformation prediction can be identified, and the prediction model and compensation strategy can be targeted.

[0142] The calculation process of the deformation deviation value takes into account various factors such as the material properties, geometric structure, and soldering process of the flexible circuit board, and can accurately reflect the difference between the deformation prediction and the actual deformation. Through continuous iterative calculation and compensation adjustment, the deformation deviation value can be gradually reduced, ultimately achieving high-precision deformation prediction and compensation. In practical applications, when the deformation deviation value is less than 0.05 mm, it can be considered that the prediction accuracy meets the requirements and no further adjustment is needed.

[0143] The method is suitable for various types of flexible circuit boards, including single-layer boards, double-layer boards and multi-layer boards. For different types of flexible circuit boards, their specific deformation characteristics can be adapted by adjusting parameters and models. The implementation effect of the method is remarkable, which can improve the deformation prediction accuracy of flexible circuit boards by more than 50%, providing strong support for the production and manufacturing of high-precision electronic products.

[0144] In an optional embodiment, a feedback compensation matrix is constructed based on the deformation deviation values, each compensation unit in the feedback compensation matrix forms a compensation transmission link according to the stress transmission direction, and the process parameters are sequentially adjusted according to the transmission characteristics of the compensation transmission link to obtain adjusted process parameters, including:

[0145] According to the distribution of the deformation deviation values, a deformation feature point is determined, a deformation coordinate system is established based on the deformation feature point, and a stress transmission direction of the deformation feature point is calculated;

[0146] In the deformation coordinate system, a compensation unit array is divided according to the stress transmission direction, the positional relationship between the deformation deviation value of each compensation unit and the deformation feature point is calculated, and a feedback compensation matrix is generated by arranging the deformation deviation values according to the positions of the compensation units;

[0147] Based on the stress distribution characteristics of the compensation units in the feedback compensation matrix, a main transmission node is determined, a compensation transmission link is established from the main transmission node, and the compensation values in the compensation transmission link are spatially reconstructed based on the positional association relationship between the compensation units to generate a compensation distribution matrix;

[0148] Based on the compensation values in the compensation distribution matrix, the compensation values of each compensation unit in the compensation unit array are adjusted, and a continuous compensation field is generated.

[0149] The spatial positional relationship of the compensation units in the compensation distribution matrix is converted into adjustment coordinates of the process parameters, the adjustment order of the process parameters in the adjustment coordinates is determined according to the transmission direction of the compensation transmission link, the compensation values in the continuous compensation field are applied to the process parameters according to the adjustment order to obtain adjusted process parameters.

[0150] As shown in Figure 3 The method comprises:

[0151] The deformation feature point is determined according to the distribution of the deformation deviation value to form the basis of the deformation coordinate system. The deformation feature point is the point with the most significant deformation on the flexible circuit board, which usually corresponds to the position of the maximum deformation deviation value. In actual application, the flexible circuit board can be scanned by a high-precision three-dimensional scanner to obtain three-dimensional coordinate data of the entire board surface, which is compared with the design model to calculate the deformation deviation value of each point. For example, a certain flexible circuit board has a maximum deformation deviation of 0.28 mm in the main chip area after welding, and the point with the maximum deformation in this area is selected as the deformation feature point. Taking the deformation feature point as the origin, the x-axis is established along the main deformation direction, the y-axis is established perpendicular to the main deformation direction, and the z-axis is established perpendicular to the xy plane, thereby constructing the deformation coordinate system.

[0152] After determining the deformation feature point, the stress transfer direction of the point needs to be calculated. The stress transfer direction refers to the main direction of stress propagation on the flexible circuit board, which is usually consistent with the main stress direction. The main stress direction of the deformation feature point can be determined by finite element analysis. In this embodiment, the main stress direction of the deformation feature point is 45°, indicating that the stress mainly transfers along this direction. Accurate determination of the stress transfer direction is crucial for subsequent division of the compensation unit and establishment of the compensation transfer link.

[0153] The compensation unit array is divided in the deformation coordinate system according to the stress transfer direction. The compensation unit is a small area on the flexible circuit board for local deformation compensation. The size and shape of the compensation unit should be determined according to the size and deformation characteristics of the flexible circuit board. In this embodiment, a square with a side length of 2 mm is used as the basic compensation unit, which is arranged along the stress transfer direction of 45° to form the compensation unit array. For a flexible circuit board with a size of 100 mm x 80 mm, about 2000 compensation units are divided. The center coordinates of each compensation unit can be represented by the deformation coordinate system, for example, the coordinates (10, 8, 0.15) represent that the compensation unit is located at a position 10 mm to the right of the deformation feature point, 8 mm above, and 0.15 mm in height.

[0154] The position relationship between the deformation deviation value of each compensation unit and the deformation feature point is calculated, including distance and angle. The distance represents the spatial distance from the compensation unit to the deformation feature point, and the angle represents the direction angle of the compensation unit relative to the deformation feature point. In this embodiment, the distance is calculated by three-dimensional Euclidean distance, and the angle is represented by the azimuth and elevation in the spherical coordinate system. For example, the compensation unit at coordinates (10, 8, 0.15) has a distance of 12.82 mm to the deformation feature point, an azimuth of 38.7°, and an elevation of 0.67°. At the same time, the deformation deviation value of the compensation unit is measured to be 0.18 mm.

[0155] The deformation deviation values are arranged according to the positions of the compensation units to generate a feedback compensation matrix, and the feedback compensation matrix is a data structure for describing the deformation distribution of the flexible circuit board, and each element corresponds to a deformation deviation value of a compensation unit. In this embodiment, the feedback compensation matrix is a two-dimensional array of 50x40, covering the entire flexible circuit board area. Each element value in the matrix represents the deformation deviation amount at the corresponding position, with units of millimeters. For example, the element value at position (5, 4) in the matrix is 0.18, indicating that the deformation deviation of the compensation unit at this position is 0.18 mm. The feedback compensation matrix intuitively reflects the deformation distribution characteristics of the flexible circuit board, providing a basis for subsequent compensation.

[0156] A main transfer node is determined based on the stress distribution characteristics of the compensation units in the feedback compensation matrix, and the main transfer node is the starting point of the compensation value transfer, and the compensation unit with the maximum stress is usually selected as the main transfer node. In this embodiment, by analyzing the stress state of each compensation unit in the feedback compensation matrix, it is found that the compensation unit near the main chip solder joint has the maximum stress, reaching 42 MPa, so this compensation unit is determined as the main transfer node. The position coordinates of the main transfer node are (8, 6, 0.22), and the deformation deviation value is 0.25 mm.

[0157] A compensation transfer link is established from the main transfer node, and the compensation transfer link describes the path through which the compensation value is transferred from the main transfer node to the surrounding. According to the stress transfer direction, the main transfer node is radiated outward, and adjacent compensation units in the stress transfer direction are connected to form multiple transfer links. In this embodiment, a total of 8 main transfer links are established, extending along the directions of 0°, 45°, 90°, 135°, 180°, 225°, 270° and 315°. Each transfer link contains 10-15 compensation units, covering the main stress area around the main transfer node. For example, the transfer link along the 45° direction contains the following compensation units: (8, 6, 0.22) -- (10, 8, 0.15) -- (12, 10, 0.12) -- (14, 12, 0.09) -- (16, 14, 0.07) -- (18, 16, 0.05) -- (20, 18, 0.04) -- (22, 20, 0.03) -- (24, 22, 0.02) -- (26, 24, 0.01).

[0158] The compensation values in the compensation transmission link are spatially reconstructed by compensating the position correlation between the compensation units, to generate a compensation distribution matrix. The position correlation includes two parameters, relative distance and relative direction, which are used to describe the spatial relationship between the compensation units. Spatial reconstruction refers to the process of adjusting the compensation values according to the position correlation, which takes into account the spatial attenuation and directional offset of the compensation values. In this embodiment, the spatial attenuation of the compensation values adopts an exponential decay model, with a decay coefficient of 0.8; the directional offset adopts a linear offset model, with an offset coefficient of 0.1 rad / mm. Through spatial reconstruction, the compensation values of each compensation unit are calculated, forming the compensation distribution matrix. For example, the compensation unit 5 mm away from the main transmission node has a compensation value of 0.25 x 0.8 5 = 0.082 mm, and a compensation directional offset of 0.1 x 5 = 0.5 rad, which is about 28.6°.

[0159] The compensation values of each compensation unit in the compensation unit array are adjusted based on the compensation values in the compensation distribution matrix, to generate a continuous compensation field. Transmission adjustment refers to the process of adjusting the compensation values of all compensation units in the array according to the compensation values of each compensation unit in the compensation transmission link. Spatial interpolation method is used to interpolate the discrete compensation values in the compensation distribution matrix to obtain the compensation values at any position, forming a continuous compensation field. In this embodiment, for the compensation units not on the transmission link, their compensation values are calculated by distance weighted average method. For example, for the compensation unit at position (15, 15, 0.1), the compensation values of the four compensation units on the transmission link around it are 0.09 mm, 0.08 mm, 0.07 mm and 0.06 mm, and the distances are 2.5 mm, 3.2 mm, 3.8 mm and 4.5 mm respectively, then the compensation value of the compensation unit is (0.09 x 2.5 + 0.08 x 3.2 + 0.07 x 3.8 + 0.06 x 4.5) / (2.5 + 3.2 + 3.8 + 4.5) = 0.072 mm. In this way, the compensation values of all positions on the flexible circuit board are calculated, forming a continuous compensation field.

[0160] The spatial position relationship of the compensation units in the compensation distribution matrix is converted into the adjustment coordinates of the process parameters, which are the positions in the welding equipment coordinate system and are used to guide the adjustment of the welding process parameters. Through coordinate transformation, the positions of the compensation units in the deformation coordinate system are converted into positions in the welding equipment coordinate system. In this embodiment, the welding equipment adopts Cartesian coordinate system, with the origin at the lower left corner of the equipment, the x-axis to the right, the y-axis upward, and the z-axis perpendicular to the xy plane. The coordinates (x, y, z) in the deformation coordinate system are converted into coordinates (X, Y, Z) in the welding equipment coordinate system through translation and rotation transformation. For example, the point (10, 8, 0.15) in the deformation coordinate system is converted into the point (25, 30, 0.15) in the welding equipment coordinate system.

[0161] According to the adjustment order of the process parameters in the adjustment coordinates determined according to the transmission direction of the compensation transmission link, the adjustment order refers to the order of adjusting the process parameters, which is usually performed in sequence from the main transmission node along the compensation transmission link. In this embodiment, the process parameters of the position corresponding to the main transmission node are adjusted first, and then adjusted in sequence along each transmission link outward. For example, for the transmission link in the 45° direction, the adjustment order is: (8, 6, 0.22) -- (10, 8, 0.15) -- (12, 10, 0.12) --... -- (26, 24, 0.01). This adjustment order ensures the continuity and consistency of stress transmission.

[0162] The compensation values in the continuous compensation field are applied to the process parameters according to the adjustment order to obtain the adjusted process parameters. The process parameters include welding temperature, pressure, time, etc., and different parameters have different effects on deformation. According to experimental data and empirical models, a mapping relationship between the compensation value and the adjustment amount of the process parameter is established. In this embodiment, the adjustment amount of the welding temperature is proportional to the compensation value, and the coefficient is 20℃ / mm; the adjustment amount of the welding pressure is inversely proportional to the compensation value, and the coefficient is -0.5MPa / mm; the adjustment amount of the welding time is proportional to the compensation value, and the coefficient is 2s / mm. For example, for a position with a compensation value of 0.1mm, the welding temperature is adjusted by -2℃, the welding pressure is adjusted by +0.05MPa, and the welding time is adjusted by -0.2s. In this way, the adjustment amount of the process parameter of all positions is calculated, and the adjusted process parameter distribution is obtained.

[0163] The adjusted process parameters are used to guide the welding process of the flexible circuit board. In the welding process, the welding conditions are accurately controlled according to the process parameter settings of different positions to achieve deformation compensation. Experimental results show that after deformation compensation using this method, the maximum deformation deviation of the flexible circuit board is reduced from 0.28mm to 0.05mm, the deformation compensation effect is significant, and the assembly requirements of high-precision electronic products are met.

[0164] This method is not only suitable for deformation compensation of flexible circuit boards, but also can be applied to the manufacturing process of other flexible electronic products. Through accurate control of the process parameters, the deformation error of the product is effectively reduced, and the assembly precision and reliability of the product are improved, which has broad application prospects.

[0165] In an alternative embodiment, a main transmission node is determined based on the stress distribution characteristics of the compensation units in the feedback compensation matrix, a compensation transmission link is established from the main transmission node, and a compensation distribution matrix is generated by spatial reconstruction of the compensation values in the compensation transmission link based on the positional association relationship between the compensation units.

[0166] stress field analysis is performed on the compensation units in the feedback compensation matrix to extract the principal stress direction and the principal stress intensity of each compensation unit;

[0167] A stress intensity field is constructed according to the distribution of the principal stress intensity in the deformation coordinate system, and a stress direction field is constructed according to the distribution of the principal stress direction in the deformation coordinate system. The stress intensity field and the stress direction field are combined to form a stress distribution feature;

[0168] The compensation unit with the maximum principal stress intensity is selected as a main transfer node based on the stress intensity field, a stress transfer region is established based on the stress direction field, a transfer axis is established in the stress direction of the main transfer node, and a transfer priority is determined based on the distance between adjacent transfer sub-regions in the stress transfer region and the transfer axis;

[0169] The transfer relationship of the compensation units is determined according to the stress distribution feature of each compensation unit in the stress transfer region, the transfer relationship is optimized based on the transfer priority, and the optimized transfer relationship is determined as a compensation transfer link;

[0170] The spatial distance and the direction angle between the compensation units are calculated based on the relative positions of the compensation units in the deformation coordinate system, the spatial distance and the direction angle are taken as position correlation parameters, the compensation values of the compensation units in the compensation transfer link are spatially reconstructed according to the position correlation parameters, and a compensation distribution matrix is generated.

[0171] The stress field analysis is performed on the compensation units in the feedback compensation matrix to extract the principal stress direction and the principal stress intensity of each compensation unit. The compensation unit can be regarded as a small area on the flexible circuit board, and its deformation characteristics are affected by the surrounding stress distribution. The stress field analysis uses the finite element method to divide the flexible circuit board into a plurality of small units and calculates the stress state of each unit. For a specific compensation unit, the principal stress direction and the principal stress intensity can be obtained by eigenvalue decomposition of the stress tensor. For example, for a compensation unit near a solder joint, the principal stress intensity is usually between 30-50 MPa, and the principal stress direction has an angle of 30°-45° with the direction of the solder joint.

[0172] After the stress field analysis is completed, the stress intensity field and the stress direction field are constructed in the deformation coordinate system. The stress intensity field describes the spatial distribution of the principal stress intensity in the deformation coordinate system, which can be represented by a three-dimensional scalar field. The stress direction field describes the spatial distribution of the principal stress direction in the deformation coordinate system, which can be represented by a three-dimensional vector field. The stress distribution characteristics are formed by combining the stress intensity field and the stress direction field, and the stress state on the flexible circuit board is comprehensively described. In practical applications, the stress intensity field of a certain typical flexible circuit board presents a "ridge" distribution, and the peak area is located near the solder joint. The intensity decreases from the center to the periphery. The stress direction field presents a radial distribution, radiating outward from the solder joint, and the direction changes with the position.

[0173] Based on the constructed stress intensity field, the compensation unit with the maximum principal stress intensity is selected as the main transfer node. In actual cases, the compensation unit near the main chip solder joint usually has the maximum principal stress intensity, reaching 45 MPa. This unit is determined as the main transfer node. The main transfer node is the starting point of the compensation value transfer, and the compensation values of subsequent compensation units will be calculated based on the association with the main transfer node.

[0174] Based on the stress direction field, the stress transfer area is established. The stress transfer area can be regarded as the path of stress propagation on the flexible circuit board, which usually extends along the principal stress direction. The transfer axis is established in the stress direction of the main transfer node, which represents the main path of stress transfer. The transfer axis usually extends from the main transfer node to the edge of the board, with a length of about 50-80 mm. The stress transfer area is divided into multiple transfer sub-areas, and the transfer priority is determined according to the distance of each transfer sub-area from the transfer axis. The closer the transfer sub-area is to the transfer axis, the higher the transfer priority. In actual implementation, the distance of the transfer sub-area from the transfer axis is divided into three levels: 0-5 mm for high priority, 5-15 mm for medium priority, and 15 mm or more for low priority.

[0175] According to the stress distribution characteristics of each compensation unit in the stress transfer area, the transfer relationship between the compensation units is determined. The transfer relationship describes how stress is transferred from one compensation unit to another, including the transfer direction and the transfer strength. The transfer direction is usually along the principal stress direction, and the transfer strength is proportional to the principal stress intensity. For two adjacent compensation units A and B, if the principal stress direction of A points to B, and the difference between their principal stress intensities is less than 10 MPa, it is considered that there is a transfer relationship between A and B.

[0176] The transmission relationship is optimized based on the transmission priority. The optimization process considers three factors: transmission priority, consistency of transmission direction, and uniformity of transmission strength. The transmission relationship of the high-priority area is preferentially retained; the transmission relationship with a transmission direction angle less than 30° with the transmission axis is preferentially retained; and the transmission relationship between compensation units with similar transmission strengths is preferentially retained. Through optimization, the most reasonable transmission relationship is screened out to form a complete compensation transmission link. In a typical case, five main transmission links are formed from the main transmission node, each link containing 8-12 compensation units, covering the main stress concentration area on the flexible circuit board.

[0177] After the compensation transmission link is determined, the spatial distance and direction angle between the compensation units are calculated based on the relative positions of the compensation units in the deformation coordinate system. The spatial distance is calculated by the difference of three-dimensional coordinates, and the direction angle is calculated by the vector angle. The spatial distance and direction angle are used as position correlation parameters to describe the spatial relationship between the compensation units. In actual cases, the spatial distance between adjacent compensation units is usually between 2-5 mm, and the direction angle is between 15°-45°. These parameters directly affect the transmission attenuation and direction deviation of the compensation value.

[0178] According to the position correlation parameters, the compensation values of each compensation unit in the compensation transmission link are spatially reconstructed to generate a compensation distribution matrix. The spatial reconstruction process considers two factors: spatial attenuation and direction deviation of the compensation value. Spatial attenuation refers to the characteristic that the compensation value decreases with the increase of spatial distance, which can be described by an exponential decay function; direction deviation refers to the characteristic that the compensation direction changes with the change of spatial position, which can be described by an angle deviation function. By considering these two factors, the compensation value of each compensation unit in the compensation transmission link can be accurately calculated.

[0179] In a specific example, the initial compensation value of the main transmission node is 0.15 mm. When transmitting outward along the transmission link, the spatial attenuation coefficient is 0.85 and the direction deviation coefficient is 0.12 rad / mm. The first level transmission unit is located 3.5 mm away from the main transmission node, and the calculated compensation value is 0.15 x 0.85 3.5 = 0.078 mm, and the compensation direction deviates from the main transmission node by 0.12 x 3.5 = 0.42 rad, which is about 24°. The second level transmission unit is located 4.2 mm away from the first level transmission unit, and the calculated compensation value is 0.078 x 0.85 4.2 = 0.035 mm, and the compensation direction deviates from the first level transmission unit by 0.12 x 4.2 = 0.50 rad, which is about 29°. In this way, the compensation values and compensation directions of all compensation units in the transmission link are calculated.

[0180] The compensation values and compensation directions of all compensation units are integrated to form a complete compensation distribution matrix, which is a three-dimensional matrix and contains the compensation amount and compensation direction of each position on the flexible circuit board. In actual application, the resolution of the compensation distribution matrix is usually 0.5 mm x 0.5 mm, covering the entire flexible circuit board area. Through interpolation calculation, the compensation value of any position can be obtained to realize continuous deformation compensation.

[0181] After the compensation distribution matrix is generated, it is used to guide the adjustment of process parameters. According to the compensation value in the compensation distribution matrix, the process parameters such as welding temperature, pressure, and time are accurately adjusted. For example, for the area with a positive compensation value, the welding temperature is appropriately reduced by 5-10℃, and the welding time is shortened by 0.5-1s; for the area with a negative compensation value, the welding temperature is appropriately increased by 5-10℃, and the welding time is extended by 0.5-1s. In this way, accurate compensation of the three-dimensional deformation of the flexible circuit board is realized.

[0182] After the flexible circuit board is compensated for deformation by using the method, the deformation error is reduced from 0.25mm to within 0.05mm, and the deformation compensation accuracy is improved by 80%. At the same time, the product yield is improved from 92% to 98.5%, significantly improving the production efficiency and product quality.

[0183] The method is not only suitable for single-layer flexible circuit boards, but also suitable for multi-layer flexible circuit boards. For multi-layer flexible circuit boards, each layer can be regarded as an independent deformation body, and the compensation transmission link and the compensation distribution matrix are established respectively, and then the interlayer influence is considered comprehensively to realize overall deformation compensation. Practice shows that the application effect of the method on double-layer flexible circuit boards is also very significant, and the deformation error is controlled within 0.08mm, meeting the assembly requirements of high-precision electronic products.

[0184] The second aspect of the embodiment of the application provides an electronic device, comprising:

[0185] a processor;

[0186] a memory for storing processor-executable instructions;

[0187] The processor is configured to call the instructions stored in the memory to execute the method described above.

[0188] The third aspect of the embodiment of the application provides a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions are executed by a processor to implement the method described above.

[0189] The present application can be a method, an apparatus, a system, and / or a computer program product. The computer program product can include a computer readable storage medium (or media) having computer readable program instructions thereon for performing various aspects of the present application.

[0190] Finally, it should be noted that the above-described embodiments are merely intended to illustrate the technical solutions of the present application, and are not intended to limit the present application; even though the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or equivalently replace some or all of the technical features thereof; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A flexible circuit board three-dimensional deformation prediction and compensation method based on deep learning, characterized in that, The method comprises the following steps: acquiring process parameters in the production process of the flexible circuit board, and establishing a process parameter-deformation relationship dataset according to the process parameters and actual three-dimensional deformation data of the flexible circuit board; constructing a variable-density grid unit along the distribution direction of the solder joints of the flexible circuit board, calculating a stress gradient matrix based on the stress deformation characteristic data of the variable-density grid unit, and constructing a deformation prediction parameter matrix based on the stress gradient matrix, which comprises: generating a variable-density grid unit along the distribution direction of the solder joints of the flexible circuit board, wherein the variable-density grid unit is set as a micro-scale grid unit in the solder joint distribution area and as a macro-scale grid unit in the non-solder joint area; acquiring first stress deformation characteristic data in the micro-scale grid unit, which represents the stress distribution and deformation trend in the solder joint area, and acquiring second stress deformation characteristic data in the macro-scale grid unit, which represents the stress distribution and deformation trend in the non-solder joint area; performing multi-scale deformation field reconstruction based on the grid density distribution of the first stress deformation characteristic data and the second stress deformation characteristic data to obtain a stress field reconstruction matrix; calculating the stress gradient of the variable-density grid unit based on the stress field reconstruction matrix, and generating a stress gradient matrix according to the distribution position of the variable-density grid unit; decoupling the stress field reconstruction matrix and the stress gradient matrix to extract stress deformation correlation characteristics, and constructing a deformation prediction parameter matrix according to the stress deformation correlation characteristics; inputting the process parameter-deformation relationship dataset into the deformation prediction parameter matrix to calculate three-dimensional deformation prediction data, comparing the actual three-dimensional deformation data with the three-dimensional deformation prediction data, and calculating a deformation deviation value; constructing a feedback compensation matrix based on the deformation deviation value, wherein each compensation unit in the feedback compensation matrix forms a compensation transmission link according to the stress transmission direction, and sequentially adjusting the process parameters according to the transmission characteristics of the compensation transmission link to obtain adjusted process parameters; performing soldering processing on the flexible circuit board according to the adjusted process parameters to compensate for the three-dimensional deformation of the flexible circuit board.

2. The method of claim 1, wherein, The method comprises the following steps: acquiring process parameters in the production process of the flexible circuit board, and establishing a process parameter-deformation relationship dataset according to the process parameters and actual three-dimensional deformation data of the flexible circuit board, which comprises: collecting process parameters of the flexible circuit board during soldering, arranging the process parameters according to the execution time sequence of the soldering process according to the collection time of the process parameters, and generating a process parameter time sequence; real-time collecting actual three-dimensional deformation data of the flexible circuit board under the action of the process parameter time sequence, and calculating the deformation gradient of the actual three-dimensional deformation data at adjacent collection time points; determining a deformation compensation direction according to the deformation gradient, constructing a process parameter change matrix by using the process parameter difference value of adjacent time sequence nodes in the process parameter time sequence, and calculating a deformation compensation coefficient based on the weight distribution of the temperature process parameter difference value and the pressure process parameter difference value in the process parameter change matrix; applying the deformation compensation direction to the deformation compensation coefficient to generate a deformation compensation amount of each time sequence node; The deformation compensation quantity is used to construct a compensation adjustment curve in a temperature process parameter correlation region and a pressure process parameter correlation region respectively, and the process parameter time sequence is compensated in a region-by-region progressive manner according to the compensation adjustment curve, and a process parameter-deformation relationship dataset is established.

3. The method of claim 1, wherein, Multi-scale deformation field reconstruction is performed based on the grid density distribution of the first stress deformation feature data and the second stress deformation feature data to obtain a stress field reconstruction matrix, including: The micro-scale grid unit and the macro-scale grid unit are divided into deformation reconstruction regions according to the density gradient size, the density change rate of the deformation reconstruction region is calculated, and a reconstruction feature vector is constructed according to the density change rate; A spiral reconstruction path is generated based on the density distribution of the reconstruction feature vector, the geometric parameters of the spatial configuration of the spiral reconstruction path match the density change process of the reconstruction feature vector, and the first stress deformation feature data and the second stress deformation feature data are density-reconstructed and mapped along the spiral reconstruction path to generate reconstruction enhancement data; A reconstruction data transmission channel is established in the deformation reconstruction region, the direction of the reconstruction data transmission channel is consistent with the density gradient direction of the reconstruction feature vector, and the reconstruction enhancement data is transmitted along the reconstruction data transmission channel to obtain transmitted reconstruction enhancement data; The transmitted reconstruction enhancement data is used to generate a stress field reconstruction matrix according to the density gradient direction of the reconstruction feature vector.

4. The method of claim 1, wherein, The process parameter-deformation relationship dataset is input into the deformation prediction parameter matrix, three-dimensional deformation prediction data is calculated, the actual three-dimensional deformation data of the flexible circuit board is compared with the three-dimensional deformation prediction data, and a deformation deviation value is calculated, including: The process parameter-deformation relationship dataset is input into the deformation prediction parameter matrix, the process parameters are quantitatively processed based on the stress transmission channel in the deformation prediction parameter matrix, and a deformation prediction flow field is generated based on the quantized process parameters in the stress transmission channel; The deformation prediction flow field is unfolded in a three-dimensional space to obtain three-dimensional deformation prediction data, and a three-dimensional deformation prediction surface is generated based on the three-dimensional deformation prediction data; A curvature constraint condition is constructed based on the deformation amount ratio between adjacent feature positions in the actual three-dimensional deformation data, and a three-dimensional deformation measured surface is generated by tensor completion of the actual three-dimensional deformation data, and the curvature constraint condition is used to limit the spatial continuous features of the actual three-dimensional deformation data; A continuous transition surface is constructed between the three-dimensional deformation prediction surface and the three-dimensional deformation measured surface, a hyperboloid interference graph is constructed based on the hyperbolic transition features of the continuous transition surface; The wave peak position and the wave valley position of the hyperboloid interference graph are extracted, and the spatial coordinate difference value of the wave peak position and the wave valley position is calculated as a deformation deviation value.

5. The method of claim 1, wherein, A feedback compensation matrix is constructed based on the deformation deviation value, each compensation unit in the feedback compensation matrix forms a compensation transmission link according to the stress transmission direction, and the process parameters are sequentially adjusted according to the transmission characteristics of the compensation transmission link to obtain adjusted process parameters, including: determining a deformation feature point according to a distribution of the deformation deviation values, establishing a deformation coordinate system with the deformation feature point, and calculating a stress transmission direction of the deformation feature point; dividing an array of compensation units in the deformation coordinate system according to the stress transmission direction, calculating a position relationship between a deformation deviation value of each compensation unit and the deformation feature point, and generating a feedback compensation matrix by arranging the deformation deviation values according to positions of the compensation units; determining a main transmission node based on a stress distribution feature of the compensation units in the feedback compensation matrix, establishing a compensation transmission link from the main transmission node, and performing spatial reconstruction on compensation values in the compensation transmission link based on a position association relationship between the compensation units to generate a compensation distribution matrix; performing transmission adjustment on compensation values of the compensation units in the array of compensation units based on the compensation values in the compensation distribution matrix, and generating a continuous compensation field; converting a spatial position relationship of the compensation units in the compensation distribution matrix into adjustment coordinates of process parameters, determining an adjustment order of the process parameters on the adjustment coordinates according to a transmission direction of the compensation transmission link, and applying the compensation values in the continuous compensation field to the process parameters according to the adjustment order to obtain adjusted process parameters.

6. The method of claim 5, wherein, determining a main transmission node based on a stress distribution feature of the compensation units in the feedback compensation matrix, establishing a compensation transmission link from the main transmission node, and performing spatial reconstruction on compensation values in the compensation transmission link based on a position association relationship between the compensation units to generate a compensation distribution matrix, including: performing stress field analysis on the compensation units in the feedback compensation matrix, and extracting a main stress direction and a main stress intensity of each compensation unit; constructing a stress intensity field according to a distribution of the main stress intensity in the deformation coordinate system, constructing a stress direction field according to a distribution of the main stress direction in the deformation coordinate system, and combining the stress intensity field and the stress direction field to form a stress distribution feature; selecting a compensation unit with the maximum main stress intensity as the main transmission node based on the stress intensity field, establishing a stress transmission region based on the stress direction field, establishing a transmission axis in the stress direction of the main transmission node, and determining a transmission priority based on distances between adjacent transmission sub-regions in the stress transmission region and the transmission axis; determining a transmission relationship of the compensation units according to stress distribution features of the compensation units in the stress transmission region, optimizing the transmission relationship based on the transmission priority, and determining the optimized transmission relationship as a compensation transmission link; calculating spatial distances and direction angles between the compensation units based on relative positions of the compensation units in the deformation coordinate system, taking the spatial distances and the direction angles as position association parameters, and performing spatial reconstruction on compensation values of the compensation units in the compensation transmission link based on the position association parameters to generate a compensation distribution matrix.

7. An electronic device, comprising: including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the method of any one of claims 1 to 6.

8. A computer-readable storage medium having stored thereon computer program instructions, wherein, The computer program instructions, when executed by the processor, implement the method of any one of claims 1 to 6. The computer program instructions, when executed by the processor, implement the method of any one of claims 1 to 6.

Images