- 學位論文
應用基因演算與電荷模擬法於最佳化絕緣礙子外型設計
Optimal Design of Insulator’s Contours Using Genetic Algorithm and Charge Simulation Method
摘要
電力系統中，為了電氣絕緣與安全，使用大量的絕緣礙子以隔絕電力與接地系統。礙子絕緣能力的不足可能導致電壓閃絡的情形發生，造成電力供應不穩定，甚至造成停電事故。絕緣能力的好壞與表面切線電場的均勻與否有莫大的關係，而電場的分布又決定於礙子本身外型輪廓，良好的礙子外型輪廓設計，將有助於提昇礙子絕緣的效果。 為達到良好的絕緣礙子外型設計，本文利用電荷模擬法分析絕緣礙子在供電情形下的電場分布，並針對支持礙子與懸垂礙子，使用不同之基因演算法最佳化絕緣礙子之外型，以期不同礙子表面切線電場能均勻並且最小化。本文在最佳化支持礙子外型設計，使用之描述外型的曲線又分指數型曲線、橢圓曲線與直線三種，同時，在支持礙子外型設計上，為避免在基因演算法在執行過程中，花費大量時間在電荷模擬法重複計算已知待選組合解之適應函數值，本文利用雜湊技術在基因演算法上，有效的存取已被電荷模擬法計算過的適應函數值，在不改變最佳化的結果下，用以提昇基因演算法的執行效能。在懸垂礙子外型設計上則使用輪廓點與三次平滑曲線代表礙子外型，再透過可根據目前已獲得之最佳解動態調整搜尋範圍之動態調整基因演算法決定最佳懸垂礙子之外型。 為驗證所提出之絕緣礙子外型最佳化設計方法，本文利用Matlab視窗軟體撰寫絕緣礙子外型最佳化程式。經由最佳化結果與原始外型比較，本文所提出之最佳化外型設計方法，不論在支持礙子或懸垂礙子上，均可獲得較為均勻分布之切線電場。在支持礙子最佳化外型設計上，雜湊技術的使用也大量減少基因演算法約92.2%的執行時間。
並列摘要
For electricity insulation and security in the power systems, insulators are widely used to isolate power lines from the grounds. Insufficient insulation of an insulator may lead to flashover, causing unreliable or even interrupted power supply. The degree of insulation of an insulator is related to distribution of tangential fields on its surface. On the other hand, the distribution of tangential fields is determined by the contours of insulator. A superior contour design of the insulator may improve the isolation effectiveness of the insulators. Aiming at achieving superior contour design of the insulator, the thesis employs the charge simulation method (CSM) to analyze the electric field distribution of the energized insulator. To obtain uniform and minimal tangential fields along the insulator surface, different genetic algorithms (GA) are used to optimize the contours of the support and the suspension insulators, respectively. Regarding the contour design of the support insulators, three different shapes of exponential, elliptic and linear functions are used to describe the contours of the support insulator. Besides, in the contour design of the support insulator, to prevent GA from spending most of its efforts in repeatedly calculating the fitness functions of the same solutions searched before through the time-consuming CSM, a Hashing technique is used in the GA to store and access the calculated fitness values of the solutions searched before. Via the usage of the Hashing table in the GA, the calculation efficiency of the GA is greatly upgraded without the expense of the solution quality. In the contour design of suspension insulator, contour points and cubic spline functions are used to describe the shape of the insulator. Based on the describing functions, the optimal contour design of the suspension insulator is achieved by using the dynamically adjustable GA, where the search ranges are adjusted successively according to the best solution found so far. To verify the proposed approaches for optimal contour design of the insulators, the methods have been implemented in Matlab programming package. Comparing the proposed optimal designs with the initial ones in terms of the distribution of tangential fields, it is shown that more uniform distributions along the insulator surfaces have been obtained from the proposed approaches for both the support and the suspension insulator. Utilization of the Hashing technique in the support insulator contour design also results in the execution time reduction of the GA up to 92.2% around.