利用MANFIS-PSO 可變慣性飛輪控制之風場調查：以東吉島為例

Wind Energy Investigation based on MANFIS-PSO Control with Inertia-tunable Flywheel: A Case Study on the Dongji Island

如何因應近百年來全球暖化造成的氣候變化已成為國際間的重要議題，各國都在積極發展再生能源。目前，再生能源發電占全台總用電量的5％，其中以風力發電為主要。然而，風能之優劣攸關風場開發效益，故如何找尋具風力潛能之開發場址為其首要之務。目前全臺灣的陸域風場已經設置了323 架風力發電機，可見陸域風場之開發已趨近飽和；由於台灣海域風能極佳，故離岸風場之推動已成為政府主要綠能政策之一。雖然海域風場較陸域來為穩定，但是也較缺乏觀測資料。為此，本研究提出一以可變慣性飛輪控制方法來穩定風力發電系統輸出，並透過複數類神經模糊網路結合粒子群演算法來最佳化PID 控制參數；透過人工智慧模擬結合局部風能資料的方式來評估潛在可開發的風能，以改善長時間觀測資料不足所可能造成的誤差，進而提供一更有效且快速的風場評估方式。

How to deal with the extreme weather resulted from global warming has become the most important issue that elevates the development of renewable energy among countries in the world. Nowadays, the renewable energy occupies at about 5% of total electricity consumption in Taiwan where wind energy is the main of it. However, the exploring efficiency of wind farm depends on the potential of wind energy. How to locate the wind farms with higher energy potential is most important. At present, 323 land-type wind power generations were established in Taiwan. That is, the exploration of land-type wind farms has been approaching to saturation. For the off-shore wind energy of Taiwan is excellent, the promotion of the off-shore wind farms has become one of the green energy policies. Although the stability of off-shore wind farms is better than the land-type ones, the previous is less of the observation data than later. In order to improve this problem, a wind energy investigation based on the MANFIS-PSO control with inertia-tunable flywheel is proposed to evaluate the potential wind energy for the possible wind farms. For the MANFIS-PSO control is the integration of multiple neuro-fuzzy networks and particle swarm organization to optimize the PID parameters, through this artificial intelligent algorithm combining with the local wind data can minimize the error occurrence of data inadequateness in long period.