Repair temperature curve setting method, system and equipment for BGA (Ball Grid Array) packaging device and medium

Abstract

The invention discloses a repair temperature curve setting method, system and equipment for a BGA (Ball Grid Array) packaging device and a medium. The method comprises the following steps: acquiring an initial data set of an SMT (Surface Mount Technology) repair link; preprocessing the temperature curve data in the initial data set of the SMT repair link to obtain preprocessed temperature curve data; using the SMT repair link initial data set and the preprocessed temperature curve data to construct a repair equivalent model based on a multi-target differential evolution algorithm, and performing iterative solution to obtain a temperature curve; based on the temperature curve, taking a heating factor as an optimization constraint, setting a repair table process parameter search range in an initial data set of an SMT repair link, synchronously optimizing the repair table process parameters based on a multi-target adaptive evolutionary algorithm of an agent model, and outputting an optimal temperature curve; the system, the equipment and the medium are used for implementing the method. According to the method, the dependence on artificial experience is effectively reduced, the debugging time is shortened, the repair efficiency and the one-time passing rate are improved, and high-precision and high-interpretability temperature curve setting is realized.

Description

Technical Field

[0001] This invention belongs to the field of BGA rework process technology, and specifically relates to a method, system, equipment and medium for setting rework temperature profiles for BGA packaged devices. Background Technology

[0002] BGA rework is essentially a complex transient heat transfer-mechanics-material coupling process: when the rework station locally heats the BGA, the solder balls, substrate, chip, molding compound, and PCB—multi-layered heterogeneous materials—undergo heating, melting, and cooling. Excessively high or uneven temperatures can lead to problems such as damage to plated vias, component cracking, and solder joint failure. Furthermore, unreasonable heating rates and holding times can reduce the mechanical and electrical reliability of the solder joints. Therefore, how to rationally set process parameters in the rework station to ensure that hundreds of solder joints undergo the required reflow profile as uniformly as possible is the core issue for guaranteeing rework quality. Current temperature profile parameter settings often rely on empirical experiments or orthogonal experiments, which are time-consuming, labor-intensive, and difficult to generalize. They also rely too heavily on human experience, lack personalization and interpretability, and are not only inefficient but also difficult to adapt to complex and changing manufacturing environments. On the one hand, some scholars have attempted to overcome these shortcomings by introducing CFD and finite element methods to model the heat transfer mechanism during the rework process. While these methods can reveal the distribution patterns of temperature fields and provide physical support for parameter optimization, they are computationally expensive and difficult to apply in real time. On the other hand, some scholars have tried to use data-driven methods to search for the optimal parameter combination, achieving better rework results. However, the model training requires a large amount of data and has high computational costs, making it difficult to apply in real time and lacking the ability to generalize to new devices.

[0003] Research on intelligent optimization of temperature profiles for BGA packaged devices in the rework stage of electronic assembly SMT production lines is scarce both domestically and internationally. Most studies on rework temperature profiles focus on modifying the rework station structure to meet the rework requirements of various devices, without addressing the core mechanisms of rework. Furthermore, because the rework stage and reflow soldering stage share some similarities in mechanism, they can serve as alternatives for comparing temperature profile optimization techniques. Therefore, this paper introduces intelligent optimization schemes for temperature profiles in the reflow soldering stage, mainly divided into two aspects: temperature profile parameter optimization based on numerical simulation and temperature profile parameter optimization based on data-driven methods.

[0004] Regarding the optimization of temperature curve parameters based on numerical simulation, Esfandyari et al. established a finite element model of reflow soldering with an overpressure module in "Simulation, optimization and experimental verification of the over-pressure reflow soldering process" (Esfandyari A, Bachy B, Raithel S, et al. Simulation, optimization and experimental verification of the over-pressure reflow soldering process[J]. Procedia Cirp, 2017, 62: 565-570.). They simulated the heat distribution using ANSYS, combined with experimentally measured temperature curves and calculated the heat transfer coefficient, analyzed the solder joint void ratio using X-rays, and optimized pressure parameters, transfer speed, etc., to clarify the optimal process conditions. However, this study did not address the compatibility with different PCB sizes and components; the mechanistic model required many parameters and relied on extensive experimental iterations. Regarding the optimization of temperature curve parameters based on data-driven methods, in "An Optimized Process for Lead-Free and Halogen-Free Reflow Soldering Based on an Artificial Neural Network Model" (Feng Zehu. An Optimized Process for Lead-Free and Halogen-Free Reflow Soldering Based on an Artificial Neural Network Model [J]. Ordnance Materials Science and Engineering, 2015, 38(4): 109-112.), Feng Zehu constructed a 6×24×2 three-layer neural network model, inputting the alloy grade and six process parameters related to preheating / welding, and outputting the push-pull force of the solder joint and the surface insulation resistance. After training with 32 sets of samples, the prediction relative error was 1.1%~3.5%. The optimized solder joint performance of the three alloys met the standards and environmental protection requirements. However, it was only adapted to three alloys and specific equipment, resulting in insufficient generalization; the training sample size was also small.

[0005] Scholars both domestically and internationally have achieved certain results in the optimization of temperature curve parameters, but limitations still exist: 1) In the optimization of temperature curve parameters based on numerical simulation, the mechanism models of BGA and PCB that need to be constructed are large in scale and have many details, and rely on a large number of experimental iterations, which makes the simulation efficiency low and may not be able to guarantee the global optimal solution, making it difficult to apply in practice. 2) Regarding temperature profile parameter optimization, while the optimization models built using neural networks and various optimization algorithms show good performance, their generalization ability is limited and they require a large amount of experimental data. This is significantly inconsistent with the small sample size in rework scenarios. Furthermore, it does not take into account the fact that rework scenarios only use a few temperature profiles, and that new devices require manual trial and error to set temperature profile parameters, thus still presenting issues with accuracy and interpretability.

[0006] As electronic products develop towards high performance, miniaturization, and multifunctionality, BGA (Block Grid Array) has become one of the mainstream forms of integrated circuit packaging due to its high I / O density, excellent electrical and thermal performance, and good reliability. Due to the characteristics of BGA chips, defects are easily generated after reflow soldering, requiring rework. BGA rework commonly uses hot air / heat baking rework stations to locally heat and reflow the components. IPC industry standards stipulate that printed circuit boards with BGA packaged devices should not be repaired more than three times; otherwise, the quality of the printed circuit board will be affected (Jia Zhongzhong. SMT Core Process Analysis and Case Studies [M]. 3rd ed. Beijing: China Industry and Information Technology Publishing Group, 2016: 90-102, 276-312.), and the rework profile is the key to determining the quality of rework.

[0007] The rework profile (preheating, isothermal, melting, and cooling stages) directly determines the quality of solder joint formation and the degree of thermal damage to the substrate and components. Improperly designed temperature profiles can easily lead to defects such as voids, cold solder joints, underfill cracking, and PCB delamination, causing second or even multiple reworks and severely impacting yield. Therefore, research into intelligent optimization technology for BGA packaged device rework temperature profiles is crucial for improving process parameter stability and rework product consistency, enhancing BGA rework capabilities, and increasing first-pass yield. This has practical application significance for the current rework process.

[0008] In summary, the following technical problems exist in the current optimization of rework process parameters: Technical Issue 1: The rework process is essentially a transient heat transfer problem. Any algorithm that analyzes material changes cannot do without material parameters such as thermal conductivity, specific heat capacity, and density of BGA. However, the datasheet only provides the external dimensions and electrical specifications, without involving material parameters. This makes it difficult to model BGA materials during the mechanism calculation process and obtain reliable results.

[0009] Technical Issue 2: Currently, the rework temperature profiles rely on only a few commonly used temperature profiles to complete the soldering process. At the same time, when a new device cannot be soldered using the historical temperature profile and the temperature profile needs to be adjusted, it is necessary to set up the profiles manually based on experience and trial and error. This is highly dependent on the experience of engineers, and there are problems with insufficient personalization and poor interpretability, resulting in low rework efficiency. Summary of the Invention

[0010] To overcome the shortcomings of the prior art, the present invention aims to provide a method, system, device, and medium for setting rework temperature profiles for BGA packaged devices. Based on a multi-objective differential evolutionary algorithm, a rework equivalent model is constructed. The rework equivalent model is iteratively solved using the rework heat conduction equation under incomplete parameters to obtain the temperature profile. Then, based on a surrogate model, a multi-objective adaptive evolutionary algorithm is used to intelligently optimize the process parameters of the rework station temperature profile to obtain the optimal rework process parameters. The present invention effectively reduces the dependence on manual experience, shortens the setup time, improves rework efficiency and first-pass yield, and achieves high-precision and highly interpretable temperature profile setting.

[0011] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for setting a rework temperature profile for BGA packaged devices includes the following steps: Step 1: Obtain the initial dataset for the SMT rework process; preprocess the temperature curve data in the initial dataset for the SMT rework process to obtain the preprocessed temperature curve data. Step 2: Using the initial dataset and preprocessed temperature curve data obtained in Step 1 for the SMT rework process, construct a rework equivalent model based on the multi-objective differential evolution algorithm; iteratively solve the rework equivalent model using the rework heat conduction equation under incomplete parameters to obtain the temperature curve. Step 3: Based on the temperature curve obtained in Step 2, with the heating factor as the optimization constraint, set the search range of the rework station process parameters in the initial dataset of the SMT rework process obtained in Step 1, and simultaneously optimize the rework station process parameters based on the surrogate model multi-objective adaptive evolutionary algorithm to output the optimal temperature curve.

[0012] The specific method of step 1 includes: Step 1.1: Obtain the initial dataset for the SMT rework process The initial dataset for the SMT rework process includes: PCB board information, BGA component information, solder paste information, rework station process parameters, rework station equipment information, and collected temperature profile data; among which: PCB board information includes: the length, width and thickness of the PCB board, the number of PCB board layers and the amount of copper on each PCB board layer; BGA device information includes: the length, width, thickness of the BGA, and whether it has a metal cap; Solder paste information includes: solder paste material and solder paste melting point; The rework station process parameters include: upper nozzle heating temperature, lower nozzle heating temperature, upper heating time, and lower heating time in three different temperature ranges; The rework station equipment information includes: the height of the heating nozzle from the component, and the solder paste material data; The collected temperature profile data includes: BGA center temperature measurement profile, BGA side temperature measurement profile, BGA side temperature measurement profile, BGA top temperature measurement profile, and PCB bottom temperature measurement profile. Step 1.2: Preprocess the temperature curve data to obtain preprocessed temperature curve data; Abnormal jump values, null values, and data points that do not conform to the actual trend are removed from the collected temperature curve data. The data is then smoothed and interpolated to obtain preprocessed temperature curve data. The BGA center temperature measurement curve in the preprocessed temperature curve data is used as the actual collected temperature curve.

[0013] The specific method for step 2 includes: Step 2.1: Using the initial dataset and preprocessed temperature curve data obtained in Step 1 for the SMT rework process, the overall properties of the BGA are preserved in an equivalent manner. An adjustment mechanism is introduced to dynamically adjust the temperature characteristic parameters of the BGA, a rework heat conduction equation is constructed, and a multi-objective differential evolution algorithm is used to optimize the equivalent parameters of the BGA to obtain the rework equivalent model. Step 2.2: Iteratively solve the equivalent model for repair using the repair heat conduction equation under incomplete parameters to obtain the temperature curve.

[0014] The specific process of step 2.1 is as follows: Step 2.1.1: Preserve the overall properties of the BGA using an equivalent approach, and introduce an adjustment mechanism to dynamically adjust the temperature characteristic parameters of the BGA: First, we make the following assumptions: We only consider heat conduction along the thickness direction of the PCB board, neglecting lateral surface heat diffusion, and treat the PCB board as a copper layer material. The thermal properties of each layer, i.e., the thermal conductivity, are considered. k Specific heat capacity c p and density ρ Uniform and consistent; The rework structure consists of three layers: BGA, solder paste layer, and PCB board. The model focuses on vertical heat conduction and the heat transfer process between these three materials. The temperature change between the top of the BGA and the bottom of the PCB board is set as the time-varying boundary condition, while density and specific heat capacity are set as temperature constants. The multiple layers of materials in the BGA are stacked and equivalent to one. The thermal conductivity of the equivalent BGA is introduced into a dynamic adjustment mechanism. This mechanism, based on a smoothing function, controls the gradual decrease in the thermal conductivity of the BGA as temperature increases, and is expressed as: in: Let be the smoothing function, and s be the slope control factor. The initial thermal conductivity is , To prevent the thermal conductivity from changing too drastically and exceeding the constraint minimum value; Step 2.1.2: Construct the heat conduction equation for repair: Let heat transfer via convection be... Q The amount of heat absorbed by the BGA and PCB board per unit time during the rework process is , h The convective heat transfer coefficient is... A For heat exchange area, For hot air temperature, For the surface temperature of BGA and PCB board, For material density, c p For specific heat capacity, V This refers to the volume of the BGA and PCB board within the soldering area. The heat input from the upper and lower air nozzles per unit time is: Heat dissipates through thermal conduction in the PCB board layers, BGA layer, and solder paste layer, following Fourier's law. The heat input per unit volume is expressed as: in: k The thermal conductivity of the material This represents the temperature gradient along the z-axis. Assuming all external heat input is absorbed, according to the definition of heat capacity, the rate of energy change per unit volume is: Combining energy change and heat conduction processes, and based on the principle of energy conservation, we establish the total energy balance equation per unit volume, namely the heat conduction equation: Step 2.1.3: Use a multi-objective differential evolution algorithm to evaluate the key parameter of the BGA: density ( ), thermal conductivity ( ) and specific heat capacity ( ), and optimize to obtain the equivalent model for rework.

[0015] The specific process of step 2.2 includes: The finite difference method was used to discretize the rework heat conduction equation over time, and the density of the BGA optimized in step 2.1 was used. ), thermal conductivity ( ) and specific heat capacity ( The time-step iterative solution of the rework heat conduction equation under incomplete parameters includes the following steps: Step 2.2.1: Using the time node of the BGA center temperature measurement curve after preprocessing in Step 1.2 as the step length, calculate the internal temperature value of the rework structure step by step. Step 2.2.2: In each iteration, the error between the calculated temperature and the temperature measurement curve at the center of the BGA is verified. If the error exceeds the preset threshold, it is fed back to the parameter optimization stage in step 2.1.3, and the population iteration strategy is adjusted and optimized again. Step 2.2.3: When the error between the iteratively calculated temperature curve and the BGA center temperature measurement curve meets the accuracy requirements, terminate the iteration and output the complete BGA center temperature curve.

[0016] The specific method for step 3 includes: Step 3.1: Based on the temperature curve obtained in Step 2, using the heating factor as the optimization constraint, set the search range of the rework station process parameters, and construct a multi-objective adaptive evolutionary algorithm based on the surrogate model: The BGA center temperature curve obtained in step 2.2.3 is optimized by setting limits on heating rate, peak temperature, time above solder joint, and heating factor as optimization constraints, with the following expressions: in: This indicates that the heating rate should not exceed 3℃ / s to prevent thermal shock damage to the device. , Temperature at peak time Temperatures above 210℃~230℃; , This indicates a temperature exceeding the melting point of solder paste. The time should be sufficient to ensure adequate welding while avoiding oxidation; among which... Represents a vector of process parameters; For the rework station process parameters in step 1.1, set the parameter search range and construct a multi-objective adaptive evolutionary algorithm based on the surrogate model for optimizing the rework station process parameters. Constructing optimization variables That is, the upper nozzle heating temperature in three different temperature ranges. Heating time Lower nozzle heating temperature and heating time Using the heating factor as the objective function, the objective function is established as follows: in: The melting point of the solder paste; Indicates the heating factor; This indicates that the welding center temperature has reached its first peak. Time; This indicates the time it takes for the temperature at the welding center to reach its peak temperature. This indicates that the welding center temperature has reached its second peak. Time; Step 3.2: Simultaneously optimize the rework station process parameters using a multi-objective adaptive evolutionary algorithm based on a surrogate model, and output the optimal temperature curve: The specific steps are as follows: Step 3.2.1, Initialization: The initial sample set is generated by uniformly sampling in the parameter space using Latin hypercube or Sobol sequences. Each sample in the initial sample set corresponds to a complete set of rework station process parameters. The samples are substituted into the rework equivalent model constructed in Step 2.1 to solve the problem and output the corresponding heating factor value, constraint satisfaction status and temperature curve characteristic parameters, forming a labeled sample dataset containing rework station process parameters, objective function and constraint information. Gaussian process regression is selected as the surrogate model. The surrogate model is trained using the labeled sample dataset to complete the initial construction of the surrogate model. Step 3.2.2: In the real simulation sample update stage of each iteration of the algorithm, high-potential samples selected based on the expected improvement index (EI) or prediction uncertainty criterion are substituted into the rework equivalent model constructed in step 2.1 to obtain their labeled data, and these new samples are included in the sample library to form an expanded sample dataset; the surrogate model is iteratively updated using an incremental learning strategy; after the surrogate model is updated, the model prediction performance is verified by the root mean square error (RMSE). After verification, it can be used for population evaluation in the next iteration. Step 3.2.3: The algorithm terminates when the maximum number of algebras is reached or the Pareto front is stable. The final optimized output is a set of Pareto optimal rework station process parameters verified by proxy assistance and real simulation. The solution that satisfies the constraints is selected as the recommended rework station process parameters and substituted into the rework equivalent model constructed in step 2.1 to output the optimal temperature curve.

[0017] In step 2.1.3, a multi-objective differential evolution algorithm is used to evaluate the key parameter of the BGA: density ( ), thermal conductivity ( ) and specific heat capacity ( The specific steps for optimizing and obtaining the equivalent rework model are as follows: Step 2.1.3.1: Set model hyperparameters: population size NP Variable factors F Crossover probability CR Given the maximum number of iterations, an initial population is randomly generated. Each individual is a target vector representing a set of candidate material components. For each individual in the population, its three objective function values ​​are calculated: density ( ), thermal conductivity ( ) and specific heat capacity ( ); Step 2.1.3.2: Define the objective function of the optimization algorithm, which is the three objective function values ​​obtained in Step 2.1.3.1: density ( ), thermal conductivity ( ) and specific heat capacity ( Substitute the temperature curve obtained by the iteration into the heat conduction equation constructed in step 2.1.2, and perform mutation, crossover and selection operations on each target vector in the population in turn. After the calculation is completed, Pareto dominance is determined and crowding is sorted. Individuals are cut off according to low and high levels to form a new population. Repeat the above mutation, crossover and selection operations until the maximum number of iterations is reached. Step 2.1.3.3: All solutions output by the model represent the optimal solutions that balance the three objective function values ​​of density, thermal conductivity and specific heat capacity. This yields the corrected equivalent model.

[0018] This invention also provides a rework temperature profile setting system for BGA packaged devices, comprising: The data acquisition and preprocessing module is used to acquire the initial dataset of the SMT rework process; and to preprocess the temperature curve data in the initial dataset of the SMT rework process to obtain the preprocessed temperature curve data. The model building and solution module is used to construct an equivalent rework model based on the initial dataset and preprocessed temperature curve data from the SMT rework process, using a multi-objective differential evolution algorithm. The equivalent rework model is then iteratively solved using the rework heat conduction equation under incomplete parameters to obtain the temperature curve. The rework station process parameter optimization module is used to set the search range of rework station process parameters in the initial dataset of the SMT rework process based on the temperature curve and with the heating factor as the optimization constraint. It then uses a multi-objective adaptive evolutionary algorithm based on a surrogate model to simultaneously optimize the rework station process parameters and output the optimal temperature curve.

[0019] The present invention also provides a rework temperature profile setting device for BGA packaged devices, comprising: Memory: A computer program that stores the above-mentioned method for setting rework temperature profiles for BGA packaged devices, and is a computer-readable device; Processor: Used to implement the rework temperature profile setting method for BGA packaged devices when executing the computer program.

[0020] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the implementation of the aforementioned method for setting a rework temperature profile for BGA packaged devices.

[0021] Compared with the prior art, the present invention has the following advantages: 1) To address the issues of large size and difficulty in identifying missing parameters in traditional numerical simulation models, this invention proposes a method for constructing a rework model based on one-dimensional transient heat conduction under incomplete parameters. This method preserves the overall properties of the BGA (Body-Gate System) in an equivalent manner and employs a multi-objective differential evolution algorithm to refine the key parameters of the equivalent model, namely density (…). ), thermal conductivity ( ) and specific heat capacity ( ), and through optimization, the heat conduction equation for BGA device rework under incomplete parameters was solved, and a temperature curve that meets the process requirements was obtained.

[0022] 2) Current rework scenarios often neglect the influence of the PCB board and its onboard components, relying solely on a few temperature profiles for soldering. Furthermore, when new components cannot be soldered using historical temperature profiles, adjustments to the temperature profiles require manual trial-and-error settings based on experience. This invention, based on the fusion of mechanism and data, proposes an intelligent recommendation method for rework station process parameters using a multi-objective adaptive evolutionary algorithm based on a surrogate model. By setting optimization constraints, the multi-objective adaptive evolutionary algorithm based on the surrogate model simultaneously optimizes the rework station process parameters, outputting the optimal temperature profile. This achieves high-precision and highly interpretable intelligent setting of temperature profile parameters for rework processes.

[0023] In summary, this invention constructs a rework equivalent model based on a multi-objective differential evolution algorithm, and iteratively solves the rework equivalent model using the heat conduction equation for BGA device rework under incomplete parameters. By setting optimization constraints and simultaneously optimizing the rework station process parameters using a multi-objective adaptive evolution algorithm based on a surrogate model, the optimal temperature curve is output. This solves the problem of model solving under incomplete parameters, realizes intelligent recommendation of rework process parameters, improves the efficiency of SMT rework, and reduces process debugging costs. Attached Figure Description

[0024] Figure 1 This is a flowchart of the implementation method of the present invention.

[0025] Figure 2 The diagram shows the heat transfer process structure before and after simplification.

[0026] Figure 3 This is a schematic diagram of a typical reflow curve for solder paste.

[0027] Figure 4(a) shows the simulation results of the temperature curve obtained by iterative optimization of the equivalent model of the BGA1 core package device in the embodiment of the present invention.

[0028] Figure 4(b) shows the simulation results of the temperature curve obtained after the optimization of the equivalent model of the BGA2 core package device in the embodiment of the present invention.

[0029] Figure 4(c) is a simulation result of the temperature curve obtained by iterative optimization of the equivalent model of the BGA3 core package device in the embodiment of the present invention.

[0030] Figure 5(a) shows the temperature curve verification results of the rework station process parameters recommended by the embodiment of the present invention with BGA1 as the core package device.

[0031] Figure 5(b) shows the temperature curve verification results of the rework station process parameters recommended in the embodiment of the present invention with BGA2 as the core package device.

[0032] Figure 5(c) shows the temperature curve verification results of the rework station process parameters recommended in the embodiment of the present invention with BGA3 as the core package device. Detailed Implementation

[0033] The technical solution adopted by the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0034] like Figure 1 As shown, a method for setting a rework temperature profile for BGA packaged devices includes the following steps: Step 1: Obtain the initial dataset for the SMT rework process; preprocess the temperature curve data in the initial dataset for the SMT rework process to obtain the preprocessed temperature curve data. The specific method of step 1 includes: Step 1.1: Obtain the initial dataset for the SMT rework process The initial dataset for the SMT rework process includes: PCB board information, BGA component information, solder paste information, rework station process parameters, rework station equipment information, and collected temperature profile data; among which: PCB board information includes: the length, width and thickness of the PCB board, the number of PCB board layers and the amount of copper on each PCB board layer; BGA device information includes: the length, width, thickness of the BGA, and whether it has a metal cap; Solder paste information includes: solder paste material and solder paste melting point; The rework station process parameters include: upper nozzle heating temperature, lower nozzle heating temperature, upper heating time, and lower heating time in three different temperature ranges; The rework station equipment information includes: the height of the heating nozzle from the component, and the solder paste material data; The collected temperature profile data includes: BGA center temperature measurement profile, BGA side temperature measurement profile, BGA side temperature measurement profile, BGA top temperature measurement profile, and PCB bottom temperature measurement profile. Step 1.2: Preprocess the temperature curve data to obtain preprocessed temperature curve data; Abnormal jump values, null values, and data points that do not conform to the actual trend are removed from the collected temperature curve data. The data is then smoothed and interpolated to obtain preprocessed temperature curve data. The BGA center temperature measurement curve in the preprocessed temperature curve data is used as the actual collected temperature curve.

[0035] Step 2: Using the initial dataset and preprocessed temperature curve data obtained in Step 1 for the SMT rework process, construct a rework equivalent model based on the multi-objective differential evolution algorithm; iteratively solve the rework equivalent model using the rework heat conduction equation under incomplete parameters to obtain the temperature curve. The specific method for step 2 includes: Step 2.1: Using the initial dataset and preprocessed temperature curve data obtained in Step 1 for the SMT rework process, the overall properties of the BGA are preserved in an equivalent manner. An adjustment mechanism is introduced to dynamically adjust the temperature characteristic parameters of the BGA, a rework heat conduction equation is constructed, and a multi-objective differential evolution algorithm is used to optimize the equivalent parameters of the BGA to obtain the rework equivalent model. The specific process of step 2.1 is as follows: Step 2.1.1: Preserve the overall properties of the BGA using an equivalent approach, and introduce an adjustment mechanism to dynamically adjust the temperature characteristic parameters of the BGA: First, we make the following assumptions: We only consider heat conduction along the thickness direction of the PCB board, neglecting lateral surface heat diffusion, and treat the PCB board as a copper layer material. The thermal properties of each layer, i.e., the thermal conductivity, are considered. k Specific heat capacity c p and density ρ Uniform and consistent; The rework structure consists of three layers: BGA, solder paste layer, and PCB board. The model focuses on vertical heat conduction and the heat transfer process between these three materials. The temperature change between the top of the BGA and the bottom of the PCB board is set as the time-varying boundary condition, while density and specific heat capacity are set as temperature constants. The multiple layers of materials in the BGA are stacked and equivalent to one. The thermal conductivity of the equivalent BGA is introduced into a dynamic adjustment mechanism. This mechanism, based on a smoothing function, controls the gradual decrease in the thermal conductivity of the BGA as temperature increases, and is expressed as: in: Let be the smoothing function, and s be the slope control factor. The initial thermal conductivity is , To prevent the thermal conductivity from changing too drastically and exceeding the constraint minimum value; Step 2.1.2: Construct the heat conduction equation for repair: Let heat transfer via convection be... Q The amount of heat absorbed by the BGA and PCB board per unit time during the rework process is , h The convective heat transfer coefficient is... A For heat exchange area, For hot air temperature, For the surface temperature of BGA and PCB board, For material density, c p For specific heat capacity, V This refers to the volume of the BGA and PCB board within the soldering area. The heat input from the upper and lower air nozzles per unit time is: Heat dissipates through thermal conduction in the PCB board layers, BGA layer, and solder paste layer, following Fourier's law. The heat input per unit volume is expressed as: in: k The thermal conductivity of the material This represents the temperature gradient along the z-axis. Assuming all external heat input is absorbed, according to the definition of heat capacity, the rate of energy change per unit volume is: Combining the above energy changes and heat conduction processes, and based on the principle of energy conservation, the total energy balance equation per unit volume, i.e., the heat conduction equation, is established: Step 2.1.3: Use a multi-objective differential evolution algorithm to evaluate the key parameter of the BGA: density ( ), thermal conductivity ( ) and specific heat capacity ( ), and optimize to obtain the rework equivalent model; the specific steps are as follows: Step 2.1.3.1: Set model hyperparameters: population size NP Variable factors F Crossover probability CR Given the maximum number of iterations, an initial population is randomly generated. Each individual is a target vector representing a set of candidate material components. For each individual in the population, its three objective function values ​​are calculated: density ( ), thermal conductivity ( ) and specific heat capacity ( ); Step 2.1.3.2: Define the objective function of the optimization algorithm, which is the three objective function values ​​obtained in Step 2.1.3.1: density ( ), thermal conductivity ( ) and specific heat capacity ( Substitute the temperature curve obtained by the iteration into the heat conduction equation constructed in step 2.1.2, and perform mutation, crossover and selection operations on each target vector in the population in turn. After the calculation is completed, Pareto dominance is determined and crowding is sorted. Individuals are cut off according to low and high levels to form a new population. Repeat the above mutation, crossover and selection operations until the maximum number of iterations is reached. Step 2.1.3.3: All solutions output by the model represent the optimal solutions that balance the three objective function values ​​of density, thermal conductivity and specific heat capacity. This yields the corrected equivalent model.

[0036] Step 2.2: Iteratively solve the equivalent model for repair using the repair heat conduction equation under incomplete parameters to obtain the temperature curve; The specific process of step 2.2 includes: The finite difference method was used to discretize the rework heat conduction equation over time, and the density of the BGA optimized in step 2.1 was used. ), thermal conductivity ( ) and specific heat capacity ( The time-step iterative solution of the rework heat conduction equation under incomplete parameters includes the following steps: Step 2.2.1: Using the time node of the BGA center temperature measurement curve after preprocessing in Step 1.2 as the step length, calculate the internal temperature value of the rework structure step by step. Step 2.2.2: In each iteration, the error between the calculated temperature and the temperature measurement curve at the center of the BGA is verified. If the error exceeds the preset threshold, it is fed back to the parameter optimization stage in step 2.1.3, and the population iteration strategy is adjusted and optimized again. Step 2.2.3: When the error between the iteratively calculated temperature curve and the BGA center temperature measurement curve meets the accuracy requirements, terminate the iteration and output the complete BGA center temperature curve.

[0037] Step 3: Based on the temperature curve obtained in Step 2, with the heating factor as the optimization constraint, set the search range of the rework station process parameters in the initial dataset of the SMT rework process obtained in Step 1, and simultaneously optimize the rework station process parameters based on the Surrogate-Assisted Multi-Objective Evolutionary Algorithm (SA-MOEA) of the surrogate model, and output the optimal temperature curve.

[0038] The specific method for step 3 includes: Step 3.1: Based on the temperature curve obtained in Step 2, using the heating factor as the optimization constraint, set the search range of the rework station process parameters, and construct a multi-objective adaptive evolutionary algorithm based on the surrogate model: The BGA center temperature curve obtained in step 2.2.3 is optimized by setting limits on heating rate, peak temperature, time above solder joint, and heating factor as optimization constraints, with the following expressions: in: This indicates that the heating rate should not exceed 3℃ / s to prevent thermal shock damage to the device. , Temperature at peak time Temperatures above 210℃~230℃; , This indicates a temperature exceeding the melting point of solder paste. The time should be sufficient to ensure adequate welding while avoiding oxidation; among which... Represents a vector of process parameters; For the rework station process parameters in step 1.1, set the parameter search range and construct a multi-objective adaptive evolutionary algorithm based on the surrogate model for optimizing the rework station process parameters. Constructing optimization variables That is, the upper nozzle heating temperature in three different temperature ranges. Heating time Lower nozzle heating temperature and heating time Using the heating factor as the objective function, the objective function is established as follows: in: The melting point of the solder paste; Indicates the heating factor; This indicates that the welding center temperature has reached its first peak. Time; This indicates the time it takes for the temperature at the welding center to reach its peak temperature. This indicates that the welding center temperature has reached its second peak. The time.

[0039] Step 3.2: Simultaneously optimize the rework station process parameters using a multi-objective adaptive evolutionary algorithm based on a surrogate model, and output the optimal temperature curve: The specific steps are as follows: Step 3.2.1, Initialization: The initial sample set is generated by uniformly sampling in the parameter space using Latin hypercube or Sobol sequences. Each sample in the initial sample set corresponds to a complete set of rework station process parameters. The samples are substituted into the rework equivalent model constructed in Step 2.1 to solve the problem and output the corresponding heating factor value, constraint satisfaction status and temperature curve characteristic parameters, forming a labeled sample dataset containing rework station process parameters, objective function and constraint information. Gaussian process regression is selected as the surrogate model. The surrogate model is trained using the labeled sample dataset to complete the initial construction of the surrogate model. Step 3.2.2: In the real simulation sample update stage of each iteration of the algorithm, high-potential samples selected based on the expected improvement index (EI) or prediction uncertainty criterion are substituted into the rework equivalent model constructed in step 2.1 to obtain their labeled data, and these new samples are included in the sample library to form an expanded sample dataset; the surrogate model is iteratively updated using an incremental learning strategy; after the surrogate model is updated, the model prediction performance is verified by the root mean square error (RMSE). After verification, it can be used for population evaluation in the next iteration. Step 3.2.3: The algorithm terminates when the maximum number of algebras is reached or the Pareto front is stable. The final optimized output is a set of Pareto optimal rework station process parameters verified by proxy assistance and real simulation. The solution that satisfies the constraints is selected as the recommended rework station process parameters and substituted into the rework equivalent model constructed in step 2.1 to output the optimal temperature curve.

[0040] Experimental Analysis The following experimental design verifies the effectiveness of this invention. BGA rework data from a domestic company was used as the experimental data. Table 1 shows the data and range used in the experiment, and Table 2 shows the temperature profile parameter settings used for soldering three types of BGAs.

[0041] Table 1 shows the structured rework data collected. Field Name Parameter value Field Name Parameter value PCB board size 160*200*2.6(mm) Upper heating temperature 220~250(℃) BGA1 size 31*31*2.9(mm) Upper heating time 20~60(s) BGA2 size 27*27*2.17(mm) Upper heating rate 3(℃ / s) BGA3 size 15*9*4.41(mm) Lower heating temperature 220~250(℃) windshield position 18 (mm) Lower heating time 20~55(s) Downwind position 183.46 (mm) Lower heating rate 3(℃ / s) Windshield Model 16~36 (mm) Infrared temperature 180(℃) Table 2 Temperature profile parameter settings used for soldering three types of BGA In addition to the data collected above, temperature change data of different BGAs during the rework process on the same PCB board were also collected as a benchmark for equivalent modeling of BGAs.

[0042] Due to the complex internal material layers of a BGA, directly measuring the thermal properties of each layer is extremely costly, and the thermal conductivity changes significantly with temperature. Figure 2 As shown, an equivalent modeling strategy is adopted: the multilayer material stack in the BGA is equivalent to a single homogeneous layer, preserving its overall thermal properties; at the same time, only the heat conduction along the thickness direction of the PCB board is considered, excluding the influence of lateral surface heat diffusion, and the PCB board is regarded as a copper layer material with uniform thermal properties, which is consistent with the model assumptions set in step 2.

[0043] To verify the feasibility of the mechanism model, the three sets of rework process parameters in Table 2 above were used as input feature parameters and substituted into the rework equivalent model constructed in this invention to obtain three sets of temperature curves. Each set of temperature curves includes the temperature curve measured by the thermocouple and the temperature curve simulated by the mechanism model.

[0044] Figure 4(a) shows the comparison between the temperature curve obtained after optimization and iteration of the equivalent model with BGA1 as the core package device and the actual temperature curve collected. Figure 4(b) shows the comparison between the temperature curve obtained after optimization and iteration of the equivalent model with BGA2 as the core package device and the actual temperature curve collected. Figure 4(c) shows the comparison between the temperature curve obtained after optimization and iteration of the equivalent model with BGA3 as the core package device and the actual temperature curve collected. In Figures 4(a), 4(b), and 4(c), the blue solid lines represent the measured temperature at the center of the BGA, and the orange dashed lines represent the temperature curves obtained by iteratively solving the revised heat conduction equation.

[0045] As shown in Figure 4(a), BGA1 has the best fitting effect with a fitting coefficient of 0.99. As shown in Figure 4(b), the fitting curve trend of BGA2 is similar to that of the actual measurement with a fitting coefficient of 0.95, which is good. As shown in Figure 4(c), BGA3 also has a good fitting effect after smoothing the actual thermocouple curve with a fitting coefficient of 0.98, which is high in accuracy and can meet the needs of practical applications.

[0046] Based on the above experiments, to verify the effectiveness of the intelligent optimization method for BGA rework process parameters that integrates mechanism and data proposed in this invention, the furnace temperature profile was optimized with the minimum heating factor of lead-free conventional solder paste as the optimization objective. The heating factor diagram is shown below. Figure 3 As shown, the black area represents the region affected by the heating factor. The heating factor is a quantitative indicator that combines temperature and time, reflecting the total heat actually absorbed during the rework process, taking into account both the temperature and the duration. A heating factor close to but not exceeding the upper or lower limits indicates that the temperature profile ensures sufficient solder wetting while avoiding excessive thermal shock, which can be considered "good." A heating factor below the lower limit or above the upper limit is considered "bad," requiring adjustment of the reflow soldering process parameters.

[0047] Figure 5(a) shows the verification results of the recommended rework station process parameters with BGA1 as the core package device. Figure 5(b) shows the verification results of the recommended rework station process parameters with BGA2 as the core package device. Figure 5(c) shows the verification results of the recommended rework station process parameters with BGA3 as the core package device. In Figures 5(a), 5(b), and 5(c), the blue curves are the simulated solder paste temperature curves under the recommended rework station parameters, the yellow dashed lines are the solder paste melting point (183°C), the vertical black solid lines are the upper and lower limits of the duration above the melting point, the second red dot is within this range and is acceptable, the green and red dashed lines are the upper and lower limits of the peak temperature, the peak temperature is within this range and is acceptable.

[0048] As can be seen from Figures 5(a), 5(b), and 5(c), the recommended temperature curve parameters can meet the actual application requirements. Compared with the actual temperature curves, in addition to meeting the actual application requirements, they also better comply with the requirements of the welding process standard document IPC-J-STD-020 and the literature (Gao Jingang, Wu Yiping, Ding Han. Optimization and control of reflow curve based on heating factor [J]. Special Equipment for Electronic Industry, 2006, (08): 23-29.), effectively improving welding efficiency and welding stability.

[0049] The key points and protection points of this invention are as follows: 1) A repair model construction method based on one-dimensional transient heat conduction under incomplete parameters is proposed. It only relies on a small number of historical repair temperature curves as input. By utilizing physical constraints (energy conservation, melting point range, etc.) and multi-objective optimization, key parameters such as thermal conductivity, specific heat capacity, and density are simultaneously inverted to ensure high similarity with the thermal response of real devices. This results in a small error between the calculated temperature curve and the actual temperature curve, meeting the needs of practical applications.

[0050] 2) A method for intelligent recommendation of rework process parameters based on a multi-objective adaptive evolutionary algorithm using a surrogate model is proposed. Based on the idea of ​​mechanism and data fusion, the heating factor is used as the core indicator. With the constraints provided by the physical model, a small number of real temperature measurement samples are introduced to complete the model correction and convergence, realize the intelligent optimization of temperature curve process parameters, and solve the problem of small sample size in rework scenarios.

[0051] This invention also provides a rework temperature profile setting system for BGA packaged devices, comprising: The data acquisition and preprocessing module is used to acquire the initial dataset of the SMT rework process in step 1; and to preprocess the temperature curve data in the initial dataset of the SMT rework process to obtain the preprocessed temperature curve data. The model building and solution module is used in step 2 to use the initial dataset and preprocessed temperature curve data of the SMT rework process obtained in step 1. Based on the multi-objective differential evolution algorithm, a rework equivalent model is built. The rework equivalent model is iteratively solved by the rework heat conduction equation under incomplete parameters to obtain the temperature curve. The rework station process parameter optimization module is used to realize the temperature curve obtained in step 2 in step 3. With the heating factor as the optimization constraint, it sets the search range of the rework station process parameters in the initial dataset of the SMT rework process obtained in step 1, and simultaneously optimizes the rework station process parameters based on the multi-objective adaptive evolutionary algorithm of the surrogate model, and outputs the optimal temperature curve.

[0052] The present invention also provides a rework temperature profile setting device for BGA packaged devices, comprising: Memory: A computer program that stores the above-mentioned method for setting rework temperature profiles for BGA packaged devices, and is a computer-readable device; Processor: Used to implement the rework temperature profile setting method for BGA packaged devices when executing the computer program.

[0053] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the implementation of the aforementioned method for setting a rework temperature profile for BGA packaged devices.

Claims

1. A method for setting rework temperature profiles for BGA packaged devices, characterized in that, Includes the following steps: Step 1: Obtain the initial dataset for the SMT rework process; The temperature curve data in the initial dataset of the SMT rework process is preprocessed to obtain the preprocessed temperature curve data. Step 2: Using the initial dataset and preprocessed temperature curve data obtained in Step 1 for the SMT rework process, construct a rework equivalent model based on the multi-objective differential evolution algorithm; iteratively solve the rework equivalent model using the rework heat conduction equation under incomplete parameters to obtain the temperature curve. Step 3: Based on the temperature curve obtained in Step 2, with the heating factor as the optimization constraint, set the search range of the rework station process parameters in the initial dataset of the SMT rework process obtained in Step 1, and simultaneously optimize the rework station process parameters based on the surrogate model multi-objective adaptive evolutionary algorithm to output the optimal temperature curve.

2. The method for setting a rework temperature profile for BGA packaged devices according to claim 1, characterized in that, The specific method of step 1 includes: Step 1.1: Obtain the initial dataset for the SMT rework process The initial dataset for the SMT rework process includes: PCB board information, BGA component information, solder paste information, rework station process parameters, rework station equipment information, and collected temperature profile data; among which: PCB board information includes: the length, width and thickness of the PCB board, the number of PCB board layers and the amount of copper on each PCB board layer; BGA device information includes: the length, width, thickness of the BGA, and whether it has a metal cap; Solder paste information includes: solder paste material and solder paste melting point; The rework station process parameters include: upper nozzle heating temperature, lower nozzle heating temperature, upper heating time, and lower heating time in three different temperature ranges; The rework station equipment information includes: the height of the heating nozzle from the component, and the solder paste material data; The collected temperature profile data includes: BGA center temperature measurement profile, BGA side temperature measurement profile, BGA side temperature measurement profile, BGA top temperature measurement profile, and PCB bottom temperature measurement profile. Step 1.2: Preprocess the temperature curve data to obtain preprocessed temperature curve data; Abnormal jump values, null values, and data points that do not conform to the actual trend are removed from the collected temperature curve data. The data is then smoothed and interpolated to obtain preprocessed temperature curve data. The BGA center temperature measurement curve in the preprocessed temperature curve data is used as the actual collected temperature curve.

3. The method for setting a rework temperature profile for BGA packaged devices according to claim 1, characterized in that, The specific method for step 2 includes: Step 2.1: Using the initial dataset and preprocessed temperature curve data obtained in Step 1 for the SMT rework process, the overall properties of the BGA are preserved in an equivalent manner. An adjustment mechanism is introduced to dynamically adjust the temperature characteristic parameters of the BGA, a rework heat conduction equation is constructed, and a multi-objective differential evolution algorithm is used to optimize the equivalent parameters of the BGA to obtain the rework equivalent model. Step 2.2: Iteratively solve the equivalent model for repair using the repair heat conduction equation under incomplete parameters to obtain the temperature curve.

4. The method for setting a rework temperature profile for BGA packaged devices according to claim 3, characterized in that, The specific process of step 2.1 is as follows: Step 2.1.1: Preserve the overall properties of the BGA using an equivalent approach, and introduce an adjustment mechanism to dynamically adjust the temperature characteristic parameters of the BGA: First, we make the following assumptions: We only consider heat conduction along the thickness direction of the PCB board, neglecting lateral surface heat diffusion, and treat the PCB board as a copper layer material. The thermal properties of each layer, i.e., the thermal conductivity, are considered. k Specific heat capacity c p and density ρ Uniform and consistent; The rework structure consists of three layers: BGA, solder paste layer, and PCB board. The model focuses on vertical heat conduction and the heat transfer process between these three materials. The temperature change between the top of the BGA and the bottom of the PCB board is set as the time-varying boundary condition, while density and specific heat capacity are set as temperature constants. The multiple layers of materials in the BGA are stacked and equivalent to one. The thermal conductivity of the equivalent BGA is introduced into a dynamic adjustment mechanism. This mechanism, based on a smoothing function, controls the gradual decrease in the thermal conductivity of the BGA as temperature increases, and is expressed as: in: Let be the smoothing function, and s be the slope control factor. The initial thermal conductivity, To prevent the thermal conductivity from changing too drastically and exceeding the constraint minimum value; Step 2.1.2: Construct the heat conduction equation for repair: Let heat transfer via convection be... Q The amount of heat absorbed by the BGA and PCB board per unit time during the rework process is , h The convective heat transfer coefficient is... A For heat exchange area, For hot air temperature, For the surface temperature of BGA and PCB board, For material density, c p For specific heat capacity, V This refers to the volume of the BGA and PCB board within the soldering area. The heat input from the upper and lower air nozzles per unit time is: Heat dissipates through thermal conduction in the PCB board layers, BGA layer, and solder paste layer, following Fourier's law. The heat input per unit volume is expressed as: in: k The thermal conductivity of the material This represents the temperature gradient along the z-axis. Assuming all external heat input is absorbed, according to the definition of heat capacity, the rate of energy change per unit volume is: Combining the above energy changes and heat conduction processes, and based on the principle of energy conservation, the total energy balance equation per unit volume, i.e., the heat conduction equation, is established: Step 2.1.3: Use a multi-objective differential evolution algorithm to evaluate the key parameter of the BGA: density ( ), thermal conductivity ( ) and specific heat capacity ( ), and optimize to obtain the equivalent model for rework.

5. The method for setting a rework temperature profile for BGA packaged devices according to claim 3, characterized in that, The specific process of step 2.2 includes: The finite difference method was used to discretize the rework heat conduction equation over time, and the density of the BGA optimized in step 2.1 was used. ), thermal conductivity ( ) and specific heat capacity ( The time-step iterative solution of the rework heat conduction equation under incomplete parameters includes the following steps: Step 2.2.1: Using the time node of the BGA center temperature measurement curve after preprocessing in Step 1.2 as the step length, calculate the internal temperature value of the rework structure step by step. Step 2.2.2: In each iteration, the error between the calculated temperature and the temperature measurement curve at the center of the BGA is verified. If the error exceeds the preset threshold, it is fed back to the parameter optimization stage in step 2.1.3, and the population iteration strategy is adjusted and optimized again. Step 2.2.3: When the error between the iteratively calculated temperature curve and the BGA center temperature measurement curve meets the accuracy requirements, terminate the iteration and output the complete BGA center temperature curve.

6. The method for setting a rework temperature profile for BGA packaged devices according to claim 1, characterized in that, The specific method for step 3 includes: Step 3.1: Based on the temperature curve obtained in Step 2, using the heating factor as the optimization constraint, set the search range of the rework station process parameters, and construct a multi-objective adaptive evolutionary algorithm based on the surrogate model: The BGA center temperature curve obtained in step 2.2.3 is optimized by setting limits on heating rate, peak temperature, time above solder joint, and heating factor as optimization constraints, with the following expressions: in: This indicates that the heating rate should not exceed 3℃ / s to prevent thermal shock damage to the device. , Temperature at peak time Temperatures above 210℃~230℃; , This indicates a temperature exceeding the melting point of solder paste. The time required should be sufficient to ensure adequate welding while avoiding oxidation; among which... Represents a vector of process parameters; For the rework station process parameters in step 1.1, set the parameter search range and construct a multi-objective adaptive evolutionary algorithm based on the surrogate model for optimizing the rework station process parameters. Constructing optimization variables That is, the upper nozzle heating temperature in three different temperature ranges. Heating time Lower nozzle heating temperature and heating time Using the heating factor as the objective function, the objective function is established as follows: in: The melting point of the solder paste; Indicates the heating factor; This indicates that the welding center temperature has reached its first peak. Time; This indicates the time it takes for the temperature at the welding center to reach its peak temperature. This indicates that the welding center temperature has reached its second peak. Time; Step 3.2: Simultaneously optimize the rework station process parameters using a multi-objective adaptive evolutionary algorithm based on a surrogate model, and output the optimal temperature curve: The specific steps are as follows: Step 3.2.1, Initialization: The initial sample set is generated by uniformly sampling in the parameter space using Latin hypercube or Sobol sequences. Each sample in the initial sample set corresponds to a complete set of rework station process parameters. The samples are substituted into the rework equivalent model constructed in Step 2.1 to solve the problem and output the corresponding heating factor value, constraint satisfaction status and temperature curve characteristic parameters, forming a labeled sample dataset containing rework station process parameters, objective function and constraint information. Gaussian process regression is selected as the surrogate model. The surrogate model is trained using the labeled sample dataset to complete the initial construction of the surrogate model. Step 3.2.2: In the real simulation sample update stage of each iteration of the algorithm, high-potential samples selected based on the expected improvement index (EI) or prediction uncertainty criterion are substituted into the rework equivalent model constructed in step 2.1 to obtain their labeled data, and these new samples are included in the sample library to form an expanded sample dataset; the surrogate model is iteratively updated using an incremental learning strategy; after the surrogate model is updated, the model prediction performance is verified by the root mean square error (RMSE). After verification, it can be used for population evaluation in the next iteration. Step 3.2.3: The algorithm terminates when the maximum number of algebras is reached or the Pareto front is stable. The final optimized output is a set of Pareto optimal rework station process parameters verified by proxy assistance and real simulation. The solution that satisfies the constraints is selected as the recommended rework station process parameters and substituted into the rework equivalent model constructed in step 2.1 to output the optimal temperature curve.

7. The method for setting a rework temperature profile for BGA packaged devices according to claim 4, characterized in that, In step 2.1.3, a multi-objective differential evolution algorithm is used to evaluate the key parameter of the BGA: density ( ), thermal conductivity ( ) and specific heat capacity ( The specific steps for optimizing and obtaining the equivalent rework model are as follows: Step 2.1.3.1: Set model hyperparameters: population size NP Variable factors F Crossover probability CR Given the maximum number of iterations, an initial population is randomly generated. Each individual is a target vector representing a set of candidate material components. For each individual in the population, its three objective function values ​​are calculated: density ( ), thermal conductivity ( ) and specific heat capacity ( ); Step 2.1.3.2: Define the objective function of the optimization algorithm, which is the three objective function values ​​obtained in Step 2.1.3.1: density ( ), thermal conductivity ( ) and specific heat capacity ( Substitute the temperature curve obtained by the iteration into the heat conduction equation constructed in step 2.1.2, and perform mutation, crossover and selection operations on each target vector in the population in turn. After the calculation is completed, Pareto dominance is determined and crowding is sorted. Individuals are cut off according to low and high levels to form a new population. Repeat the above mutation, crossover and selection operations until the maximum number of iterations is reached. Step 2.1.3.3: All solutions output by the model represent the optimal solutions that balance the three objective function values ​​of density, thermal conductivity and specific heat capacity. This yields the corrected equivalent model.

8. A rework temperature profile setting system for BGA packaged devices based on the method of claim 1, characterized in that, include: The data acquisition and preprocessing module is used to acquire the initial dataset for the SMT rework process; The temperature curve data in the initial dataset of the SMT rework process is preprocessed to obtain the preprocessed temperature curve data. The model building and solution module is used to construct an equivalent rework model based on the initial dataset and preprocessed temperature curve data from the SMT rework process, using a multi-objective differential evolution algorithm. The equivalent rework model is then iteratively solved using the rework heat conduction equation under incomplete parameters to obtain the temperature curve. The rework station process parameter optimization module is used to set the search range of rework station process parameters in the initial dataset of the SMT rework process based on the temperature curve and with the heating factor as the optimization constraint. It then uses a multi-objective adaptive evolutionary algorithm based on a surrogate model to simultaneously optimize the rework station process parameters and output the optimal temperature curve.

9. A rework temperature profile setting device for BGA packaged devices, characterized in that, include: Memory: A computer program storing a rework temperature profile setting method for BGA packaged devices as described in any one of claims 1-7, and is a computer-readable device; Processor: Used to implement the rework temperature profile setting method for BGA packaged devices as described in any one of claims 1-7 when executing the computer program.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the implementation of the rework temperature profile setting method for BGA packaged devices as described in any one of claims 1-7.

Images