饋線重構是配電管理系統中用以處理問題的一項重要技術,其主要的功用是在滿足輻射狀配電架構、電壓與電流等限制條件的前提下,於正常運行情況,一方面平衡饋線負載、消除過載,提高供電電壓品質,另一方面,降低線路損失,提高系統經濟性;在緊急情況下,隔離故障,縮小停電範圍,並在故障排除後,迅速恢復供電。所以饋線重構乃是提高系統安全性和經濟性的重要手段。執行饋線重構的方式是藉由確定饋線上開關對之間的開啟、閉合狀態使得配電系統的運轉處於最佳的狀態。因此,饋線重構為一典型的非線性多目標組合最佳化問題,使用常規的數學最佳化方法難以求解。況且實際配電系統饋線上的開關數量相當龐大,可行之開關操作策略解空間是相當龐大的,要如何提升有效之開關對組合的搜尋,加速演算法的收斂速度,並維持開關操作策略解的精確度,使演算法的結果能夠逼近或達到真正的最佳解,乃為開發配電饋線重構演算法的主要工作。為達成此目標,本論文採用粒子群演算法為核心,除探討粒子群演算法應用於饋線重構問題上的可行性,並檢視多目標饋線重構問題的需求,期望能夠發展一使用柏拉圖最佳化技術之多目標粒子群演算法,有效求解多目標饋線重構問題。
Feeder reconfiguration is an important function in Distribution Management System. Feeder reconfiguration is performed by opening/closing of closed switches and opened switches. During normal distribution system operations, by changing the on/off status of these switches can reduce system loss or operate the system more reliable and economic. When the status of switch is changed, the resulting topology of the distribution system must be maintained in radial structure. The constraints such as feeder loading and voltage profile should not be violated after reconfiguration. Under abnormal condition, feeder reconfiguration can be used to restore services to de-energized zones after fault is isolated. Since the reconfiguration is done by changing the status of switches, it can be categorized as a non-linear multi-objective combinational optimization problem. It is a difficult task to obtain the best switch operation plan by applying regular mathematics optimization methods. Since there are a lot of switches on distribution system, the goal to develop an optimal reconfiguration algorithm is to eliminate unnecessary searching for switch pairs. By doing so, searching time of the algorithm can be reduced while the accuracy of results identified by the algorithm should be maintained. To achieve this goal, this dissertation adopts the particle swarm optimization for feeder reconfiguration problems. In order to develop a feasible multi-objective particle swarm optimization algorithm which embedded Pareto Optimality technique for solving the feeder reconfiguration problem, this dissertation not only discusses the feasibility of applying particle swarm optimization for feeder reconfiguration but also reviews the requirements of the multi-objective feeder reconfiguration problems.