Constraint Satisfaction Method for Switching Operation in Restoration of Distribution System

• Main Authors: Kun-Yi Lee, 李坤益
• Other Authors: Hong-Tzer Yang
• Format: Others
• Language: zh-TW
• Published: 1999
• Online Access: http://ndltd.ncl.edu.tw/handle/91212703265747055116

碩士 === 中原大學 === 電機工程學系 === 87 === Due to the complicated combinatorial problem of distribution-system restoration with on/off switching operations, it is difficult to use conventional mathematical planning methods to solve this problem. Traditionally, through a decision-tree formulation, the problem is solved by means of a searching algorithm. In the switching operation problem of a distribution system, a switch in the system will be in on or off state. Therefore, in the worst case, an n-switch system would have a complete problem space of 2n states. For a combination of restoration switching operations in a practical distribution system, the dimension of problem is often too huge to solve in a limited time. In this thesis, an application of constraint-satisfaction method is proposed to cope with the distribution-system restoration problem. By using different constraint conditions, domains of decision variables (on/off states of switches) that are not consistent with each other are deleted. Such an approach can thus greatly reduce the search space. The constraint-satisfaction method proposed in this thesis contains back-tracking operation, forward checking operation, and look-ahead inference operation. Among the feasible domains of individual decision variables, these operations find the feasible solutions according to the constraints between the decision variables. When any variable is assigned a value, the constraint-satisfaction method would employ constraints to delete the infeasible solutions from the searching space. In the searching process, the constraints are thus relied on to trim the solution space as much as possible. After that, a depth-first branch-and-bound optimization approach serves to find the optimal solution from the reduced feasible solution space. Cases of single-fault point and double-fault point in two distribution-system restoration problems are used to test the proposed constraint-satisfaction method in comparison with the traditional simulated-annealing (SA) method. The impacts of different solution methods on the results are analyzed. The testing results reveal that the constraint-satisfaction approach can find the near-optimal solution obtained from the SA method within a far shorter solution time, owing to the searching space greatly reduced by the rigorous constraints.