| 摘要: | 近年來，環保意識抬頭，全球皆積極研發使用潔淨的再生能源，減輕傳?發電方式所產生之汙染問題。使得太陽能產業得以被重視，也成為未來能源的趨勢。也因這股新趨勢，各家廠商看中這塊市場，也紛紛大批投入生產太陽能產品，又由於各國政府近年來將補助案紛紛調降，使得太陽能模組在市場上呈現一片供過於求的現象，因此有許多業者為了能消化庫存，便削價拋售，使得市場價格一路往下滑落，很多製造廠也因此倒閉。在標準型太陽能模組技術趨於成熟的時代中，有另一片市場正在崛起，那就是建材一體型太陽能模組，它能與建築物結合使用，同時達到環保與建築美觀的效果，因此本研究將針對建材一體型太陽能模組之製程參數利用資料探勘技術中的類神經網路、線性回歸、廣義線性模型、分類與回歸樹等模型來尋找影響良率的重要因子，進而提升製程良率、減少成本的支出，生產出品質優良的產品，提升自身在這競爭激烈的環境中有別於技術成熟的傳統型太陽能模組之能力。
In recent years, Environmental Protection turns to a hot-issue from a concept the world are actively developing new and clean renewable energy, to reduce the pollutions arising in traditional power generation.Therefore, it, Solar energy industry been taken seriously, moreover, it becomes a trend of the future energy. Also because of this new trend, many manufacturers fancy piece of the market and have a large number of inputs to the production of solar products, and also, governments have cut the subsidy case in recent years,
therefore, lots of firms lowering their prices in order to digest inventory, then turns the market prices fall all the way down, many manufacturers also go down the tubes. Nowadays, when the standard solar module technology is getting mature, there is another solar module market is rising, it is " Building Integrated Photovoltaics Module", it can be used as combination on building, and also benefit to the aesthetic and environmental protection. Therefore, this study will focus on the process of building integrated solar module parameters,applying the data mining technology of neural networks, linear regression, generalized linear models, classification and regression tree model to find the yield-limiting factor,and thus improve the yield, reduce their costs, and produce quality products, to enhance the ability to mature technology in this competitive environment, which is big difference from the traditional solar module. |
