Summary: | 博士 === 國立臺灣科技大學 === 電機工程系 === 104 === This dissertation presents an efficient way of solving distribution system reconfiguration (DSR) and restoration problem in electrical power systems with consideration of different type of distributed generators (DGs). The objective of the reconfiguration problem is to minimize the distribution power loss under normal operating conditions, while the restoration problem aims to simultaneously optimize power loss reduction and power delivery maximization after part of the network is isolated due to single or multiple line faults. Several algorithms have been developed in literature; however, some of them result in sub-optimal solutions while the others cost higher computational time. In this dissertation, two new DSR algorithms based on mixed integer linear programming (MILP) and a modified particle swarm optimization (PSO) are proposed. The proposed MILP based DSR algorithm reformulates the reconfiguration problem in such a way that the approximation error between the MILP model and the true non-linear model is minimized. On the other hand, the proposed modified PSO implements a number of modifications to improve the conventional meta-heuristic DSR algorithms for avoiding local optimum and reducing the size of the searching space. It also easily incorporates DGs with constant voltage control mode and integrates hourly DSR with optimal DG active power scheduling. The proposed MILP based method can ensure a global optimum solution. However, the system has to be linearised. Thus, it results in approximated solutions. On the other hand, the proposed modified PSO can provide exact solutions since linearisation is not required. Furthermore, a restoration algorithm has been developed to restore distribution systems with optimal load shedding and minimum power loss considering islanding that might occur during fault appearance. The validity and the effectiveness of the proposed methodologies have been tested using standard IEEE 33 and 69-bus networks with various case studies.
|