建構台灣地區太陽能發電系統之發電量預測模型

Abstract

電在人類生活中扮演著不可或缺的角色，而電能大多必須藉由化石燃料或鈾礦進行能量轉換所產生，這些資源在地球上的蘊藏量有限且分佈不均，且能源轉換為電力的過程中，會對於地球環境造成不少負面影響，有鑑於上述能源短缺與環保問題，太陽能、風力等再生能源發電的重要性逐漸提高，許多國家皆積極的推廣設置太陽能發電系統。然而，太陽能發電系統之轉換效率始終存在技術瓶頸，台灣雖位於太陽光能輻射豐富的地區，實際投入太陽能發電市場的廠商卻不如預期，主要的原因就是發電轉售的盈收不敷高昂的設置成本。因此，本研究分析台灣地區現有的太陽能發電系統資料，並建構一套發電量之預測模型，由於系統實際輸出的發電量，牽涉到地表日照量與系統轉換效率兩個因素，故本研究實際上是分別針對此兩因素建構預測模型，在系統轉換效率的部份，以各種系統規格做為解釋變數，應用遺傳規劃法（Genetic Programming, GP）進行預測；在地表日照量的部份，本研究搜集中央氣象局出版的氣候年報，以時間序列分析（Time Series Analysis）之成份分解法預測台灣各地區未來的地表日照量，最後整合地表日照量預測值、系統模組面積與轉換效率預測值，即可預測系統的發電量，且預測模型可程式化，只要輸入發電系統的關鍵變數與設置地區，即可得知系統在未來某一段時間內，正常運作下可輸出的發電量，一旦產、官、學各界對於太陽能發電系統輸出的發電量能有更準確的預估，對於投資、政策制定及相關學術研究皆能有不小的幫助。

Electricity plays an important role in civilized life, but most of the electricity must be generated from Fossil fuels and Uranium, these resources in the surface of the earth are limited and unevenly distributed. In addition, the process of electricity generated will cause a lot of negative impact on earth's environment. Considering the crisis of energy shortage and environmental protection, the importance of renewable energy, such as solar energy and wind power, are increasing gradually, that is why many countries are actively setting up the photovoltaic (PV) system. However, the conversion efficiency of PV system is limited to Engineering technique, although Taiwan is located in a region of abundant solar radiation, the actual investment of PV system market is not as good as expected, the main reason is that the benefit of electricity resale is inadequate for the high set-up cost. Therefore, this study analyze the PV system data now in Taiwan to constructs a predicted model for its electricity, but generated electric energy involves two factors – solar radiation and conversion efficiency, so this study actually constructs a predicted model for these two factors respectively, we apply Genetic Programming (GP) for the conversion efficiency predicted model, system specifications means the explanatory variables and conversion efficiency is the response variable; in the solar radiation predicted model, we collected the solar radiation data from annual report which published by central weather bureau (CWB), and then construct a Time-Series predicted model for the solar radiation around Taiwan by using component decomposition method. Finally, by integrating the predicted value of solar radiation, the size of PV module and the predicted value of conversion efficiency, we could evaluate the electric power which PV system can generate. In this study, the predicted model is programmable, enters the key variables of PV system and where is the site of system, and you will get the estimated value of electricity. Once the industry, the official and academics can evaluate the generated electric energy of PV system more precisely, it will be helpful for investment, policy development and related researches.