Parameter Optimization for Microlens Arrays Fabrication Using Genetic Algorithms

應用基因演算法於微透鏡陣列製程最佳化

邱垂昱(Chui-Yu Chiu) ； 林儀(Yi Lin) ； 周宜賢(Yi-Hsian Chou) ； 施柏州(Po-Chou Shih)

《中國機械工程學刊》 34卷6期 (2013/12) Pp. 507-517

摘要

微透鏡陣列是人類仿生之具體表現，是微光學元件在日常生活中之應用。其常被應用的領域包括平面顯示器的背光擴散及增效、光纖耦合，以及波前檢測等。許多製程被開發使用以提升其精密度，但能將之以較低成本大量複製的方法並不多見。本研究採用微射出成型製程實驗所得的數據進行探討。以較少的實驗結果，利用倒傳遞類神經網路建構一個預測模式，再以基因演算法進行全域最佳解搜尋。藉由上述的流程，求解最佳的製程參數組合，以最佳化製程提升微透鏡陣列之轉寫性特性。

並列摘要

As a work of bionics, microlens arrays are an application of micro-optics, which are used in daily life. They are commonly applied in the backlight diffusion and enhancement of flat panel display backlights, optical fiber couplings, and wave-front detection. Numerous processes have been developed to enhance the precision, but few of them can be applied in mass production. In this study, the experimental data of micro-injection molding process were used for exploration and discussion. Two algorithms, back-propagation neural network (BPN) and genetic algorithm (GA), are utilized to solve continuous problems with a few discrete data. This study used the BPN algorithm to construct a prediction model with a few experimental results, and then applied the GA algorithm in the global optimum search. The above process was used to find the optimal process parameter combination to assist the enhancement of the optical replication of microstructure height.

並列關鍵字

Microlens Array ； Optimization Process ； Genetic Algorithm ； BPN