在機構設計遇到應力或變形問題,常常要等產品開發週期後半段才發現,使用傳統的修改與試模的試誤法,較耗時間與經費。本研究結合實驗規劃法、類神經網路、有限元素分析與最佳化的方法,建構一套電腦輔助設計程序。透過商用軟體 ANASYS的分析功能,並結合類神經網路及基因演算法進行按鍵開關尺寸最佳化設計,試圖由產品設計限制中找出尺寸最佳形狀。本論文先運用實驗規劃法結合商用有限元素法軟體ANSYS進行結構分析,以類神經網路建構應力的數學預測模式,接著用數學預測模式及基因演算法進行按鍵開關尺寸最佳化。由結果顯示,本論文所使用的方法能快速及精準地找出按鍵開關尺寸最佳化設 計,降低產品錯誤率與節省成本。
In developing the product, the problems of stress and deformation will be occured at the late state of design cycle. Using the traditional trial-and-error procedures, design modification and mould testing, it will cost lots of time and buget to finish the job. This study integrates the method of design of experiment, neural networks, finite element analysis, and optimization algorithm for shape design optimization. Using commercial finite element program ansys for structural analysis, the mathematical model between shape and stress, meta-model, can be built by combination of design of experiment and neural network. Then the genetic algorithms are used for shape optimization based on the meta-model of shape and stress of push-button of wireless ear phone. It is shown from the results that the method developed in this study can be fast and precisely find the optimum shape with minimum stress. The cost and error can be also reduced using this approach.