应用遗传算法的线路板拆解加热参数优化  

Optimization of PCB disassembly heating parameters based on genetic algorithm

作　　者：高鹏[1] 向东[1] 杨继平[1] 程杨[1] 段广洪[1] 丁晓宇[1] 

机构地区：[1]清华大学精密仪器与机械学系,北京100084

出　　处：《现代制造工程》2008年第8期92-95,120,共5页Modern Manufacturing Engineering

基　　金：国家科技支撑计划项目(2006BAF02A16);重庆市国内科技合作资助项目(CSTC,2007AC2063)

摘　　要：为了优化线路板拆解加热参数,保证元器件有较高的拆除率和合格率,通过正交试验和BP神经网络技术,建立了加热参数与元器件拆除率和合格率之间的映射关系,并验证了网络的稳定性和可靠性;用遗传算法对无数学表达式、多目标、离散值的线路板拆解问题进行多次优化,均收敛到同一个解,说明最优解的稳定性与正确性;按照优化参数得到的元器件拆除率和合格率的网络预测值和试验值都很高,说明加热参数优化效果良好。In order to optimize heating parameters to make sure of high disassembly and pass rate of PCB components, mapping relationship between hating parameters and disassembly and pass rate was built based on orthogonal experiments and BP neural network. Stability and reliability of neural network were proved. No-expression, multi-target and discrete disassembly problem was optimized applying genetic algorithm, which converged to a stable and valid solution. Optimized parameters have high network prediction and experiment values of disassembly and pass rate.

关 键 词：线路板拆解 遗传算法 正交试验 神经网络 

分 类 号：X705[环境科学与工程—环境工程]