| 摘要: | 在一個網路發達的時代資料隨處可得，但是沒有經過整理的資料形同垃圾。要如何將靜態的、具有潛在價值的資料，有系統的變成動態的、顯見有用的資訊，又進一步挖掘並轉換成可用的知識，是非常重要的，這也就是資料探勘的精神所在。因為需要直射日照，大型聚光型太陽光發電系統多半有數十或百來座太陽光追蹤器，承載千百個聚光型太陽電池模組以追蹤太陽。一般來說，聚光型太陽光發電系統直流發電量下降的原因，可能是因為設備故障 (如：元組件劣化、損壞)、調校問題、或環境因素 (如：灰塵、花粉、油汙等) 污染造成。而設備故障、調校問題與環境因素造成系統功能的下降，在維護聚光型太陽光發電系統發電效能的方法及策略是不同的。也就是說，設備故障、調校問題只要換新或調校即可修復，但環境因素汙染造成的功能衰退，會因下雨、安裝地點或季節性而異。本研究利用迴歸分析灰塵及髒汙對太陽能發電系統影響，並透過迴歸模型解釋能力解析聚光型太陽電池模組於下雨或清洗前後的變化，同時了解到因為安裝位置的不同，太陽電池模組受環境因素影響的程度也不同，進而找尋改進的清洗策略，以節省人力、成本、及資源，更可提升系統使用度，維持最佳的太陽光發電系統之整體效益。Information is useless if it is not trimmed, sorted, or organized in nowadays the network era. Converting tacit information with potential value to explicit knowledge is not only a process of data management, but also the spirits of data mining technologies. Owing to requiring direct normal irradiance (DNI), the concentration photovoltaic (CPV) system comprises hundreds of solar trackers to carry modules to trace the sun. General speaking, possible reasons for the output power of the CPV system going down are deteriorating or failed components, adjustment problems, or under influences of environmental factors, such as: dirt, pollen, or soiling. However, maintenance methods and strategies are varied among those reasons, i.e., system will be recovered by repairing the deteriorated, failed, or making an adjustment, but to resume the output power from soiling depends on the sites, seasons, and rains. In this study, via regression technology, authors exploited the influences on CPV system by soiling or dirt; also, through R-square, to understand the trend after raining or module cleaning. In the meantime, comprehending influences on modules are varied in terms of environmental factors on different locations. In the end, a cleaning decision making strategy is promoted to save manpower, deduct resource waste, achieve cost reduction purpose, enhance system availability, and secure the best overall benefits of CPV systems. |
