| Summary: | 碩士 === 國立臺北科技大學 === 自動化科技研究所 === 100 === In recent years, the issue of resource crises and the promotion of awareness on environment protection have
driven the upsurge of solar cell industry in Taiwan. In the production process of solar cell, etch rate and performance of the cell is complementary to each other. In order to well control the variation of etch rate and simultaneously avoid waste of cost, it is a must to construct a prediction model for etch rate. The model of grey prediction can generate excellent results thanks to its advantages that it requires little data and its operation is simple. This article employed the characteristic of periodic oscillation in etch rate to enter traditional grey theory into Fourier residual error correction to further boost accuracy of prediction, and adopted the pattern of fuzzy decision to adjust the variation of etch rate caused by manually acid replenishing. During prediction, the background values were optimized to allow the best performance in prediction. In comparison with Markov residual error correction, the result shows that Fourier residual error correction created better prediction result. This study accurately predicted etch rate based on the etch rate yielded during production, which allows the process engineer to adjust process parameters accordingly and displays the advantage in effectively reducing time required for machine recovery and costs.
|
|---|
