本文針對輸電系統作最佳無效功率調度為研究目標。期能輔助調度人員,使能作更正確、經濟、安全之調度。無效功率調度是關係重大且困難執行的事,目前調度上完全需仰賴調度人員專業之經驗,以電壓過高或過低附近之並聯補償器來作調度。由於近年來地下電纜大量的被使用,第六輸電線路加入後,動態與穩態之情形與以前不同,因此針對輸電系統來作最佳化無效功率調度,以輔助調度人員實為刻不容緩之事。本文提出利用基因法則對輸電系統作最佳無效功率調度,並使用電力潮流分析得到各匯流排之電壓,以使電壓在合理範圍,避免電壓崩潰。接著利用類神經網路學習經由基因法則所得到之調度容量,作為倒傳遞網路的輸入資料依據。為了縮短倒傳遞類神經網路的訓練時間和反應時間,本文利用具有資料壓縮功用的主成份分析法來作為倒傳遞類神經網路的資料先處理,預先對輸入資料進行簡化,以減少輸入變數的數量,進而提高執行速度。本文利用一大型輸電系統進行研究,且獲得不錯之結果。
This thesis aims to do the optimal reactive power dispatch of transmission systems. The purpose is to assist dispatch persons to do more correct, economic, and secure dispatch. The dispatch of reactive power is important and difficult to do. Nowadays, the dispatches are almost from operators' experiences. The compensators of over-voltage or under-voltage are used. However, the large amount of underground cables are built to power systems, the dynamic and steady-state stability is different from those before the Sixth transmission systems. The assistant from optimal reactive power dispatch is an important thing. The assistant from optimal reactive power dispatch is an important thing. This paper is proposed genetic algorithm to do an optimal shunt capacity dispatch of a transmission system. To modify the bus voltage to reasonable margin, avoid the risk of a voltage collapse. For the dispatch, the artificial neural network is applied to learn the dispatches from optimal methods by GA, a data-reduction skill was used by principal component analysis. It can decrease the number of input and increase the speed of the artificial neural network. Simulation results show that the proposed algorithm obtains the good performances on the transmission system.