因為再生能源在近年來的發展，更全面的節能以及永續發展並有效的提高能源使用效率對於發展電力系統及智慧電網是很重要的，也因此電力調度的需求也逐漸地提高，近年來政府持續推動如綠能屋頂等綠能供電設備的安裝讓家庭得以自主發電並減少市電的使用並搭配儲能系統就可將一個家庭視為一個小型微電網，當這樣的家庭組成社區並以不加入市電的前提下長久運作就形成了多重微電網系統，電力調度的技術能讓多個電網能互相溝通、輸送電力，因為系統中的電網能吸收多餘的電力，而且還可以互相支持作為備用電源進而讓能源的利用率達到最佳化，但傳統的多電網電力調度是以中央集中管理控制的系統來實現，將各電網控制器的數值匯集至中央控制器集中優化後再將命令回傳至各電網控制器，儘管這樣的控制系統可以得到所有電網的資訊並處理，但也因此造成中央系統有巨大的運算與通訊負擔，並且運用於以不加入市電為前提的孤島多重微電網系統而言可能會因集中運算的調度命令不夠即時而導致負載被迫中斷或因此損壞，因此本論文將建立以去集中化控制為基礎的多微電網控制系統並加入電網運作成本為計算參數以獲得最佳電力調度。 本文模擬採用美國國家儀器公司所開發之模擬軟體Labview來模擬本文電力調度策略的可行性，並使用Python撰寫相關類神經演算法，最終驗證演算法之結果在調度成本的優化與系統可靠性。 ;In recent years, under the development of renewable energy, the demand for power dispatch has gradually increased. More comprehensive energy use and sustainable development. Improving energy efficiency is very important for the development of power systems and smart grids. In recent years, the government has continued to promote installation of green energy power generation such as energy roofs allows families to generate electricity independently and reduce the use of commercial power. When combined with an energy storage system, a family can be regarded as a small microgrid. When such families form a community and sharing their energy. The community can be regarded as a multi-microgrid system. The power dispatching technology allows multiple grids to communicate with each other and transmit power, because the grids in the system can absorb excess power, and can also support each other as a backup power source for energy utilization However, the traditional multi-microgrid power dispatch is realized by a centralized control system. The data of the grid are collected to the central controller for centralized optimization and then the commands are transmitted back to the grid controllers. Although such the control system can obtain and process information from all microgrid, it also causes a huge computational and communication burden on the central system. If the dispatch command of the calculation is not instant enough, the load may be interrupted or damaged. Therefore, this paper will establish a multi-microgrid system based on decentralized control and add the regulation cost as a calculation parameter to obtain the best power dispatch. This paper uses the simulation software Labview to simulate the feasibility of the power dispatching strategy, and uses Python to write related neural algorithms, and finally verify the results of the algorithm in the optimization of dispatching cost and system reliability.
