Power system restoration by joint usage of expert system and mathematical programming approach

When electric power supply interruption is caused by a fault, it is imperative to restore the power system promptly to an optimal target configuration after the fault. The problem of obtaining a target system is referred to as power system restoration. Both mathematical programming (MP) and expert systems (ES) have been used to solve restoration problems. However, existing approaches based on either MP and ES alone have inherent limitations stemming from their own paradigms. Mathematical progressing can obtain an optimal configuration under specified operational constraints, but requires a relatively long solution time. Although ES are effective in that they can utilize expert knowledge, maintenance of large-scale ES requires inordinate effort. This paper proposes a new approach to power system restoration that utilizes both methodologies so as to exploit both systems' advantages. That is, a system under study is decomposed into a set of subsystems based on the knowledge of restoration experts, which is realized as an expert system. Then, MP is applied to each decomposed subsystem to obtain an optimal target configuration. A feasible operation sequence leading to the target configuration is generated by ES. This approach reduces significantly the computation time required to obtain target systems and is far less than would be the case if the total system is solved (as a unity). Moreover, the number of rules in the knowledge-base are greatly decreased.