摘要
回流焊温度曲线的输出特性是衡量SMT生产过程质量的重要指标。理想的输出曲线往往需要反复多次地调试输入参数才能获得,针对输入参数难以快速设定的问题,采用一种模糊联想记忆神经网络来描述回流焊过程的输入参数与温度曲线的输出特性参数的动态、非线性关系,利用fuzzyTECH模糊软件搭建模糊推论平台,以历史数据和工程师的操作经验为依据建立模糊规则库,然后以理想输出结果为依据,通过模糊推论算法来获得最佳输入参数。结果表明,该方法较试误法提高了参数设定的精确度,并显著缩短了回流焊的参数设定时间。
The characteristic of reflow soldering profile is the most important index in measuring the quality of SMT production process.The expected output profile was usually obtained through numerous adjustments of the input parameters.Input parameters could be set more quickly by adopting a fuzzy associative memory neural networks,which describes the dynamic and nonlinear relationship between the input parameters of the reflow soldering and the output characteristic of the profile,using historical data and operational experience to build the fuzzy rule block through fuzzyTECH,and at last getting the ideal input parameters＇ value through FAM approach based on the expected output.The result shows that this approach,compared with the trial-and-error method,improves accuracy and reduces the parameter-setting time effectively.
出处
《工业工程与管理》
北大核心
2010年第6期75-81,共7页
Industrial Engineering and Management
基金
国家自然科学基金项目(50875168)
关键词
模糊联想记忆神经网络
回流焊
模糊规则库
语言变量
fuzzy associative memory neural network
reflow soldering
fuzzy rules
linguistic variables
