A Case Learning System for Switching Instruction Tables in Load Dispatching Offices

Abstract

The authors have proposed a new framework of automatic knowledge acquision method and its prototype system has been reported at⁽⁶⁾, which has generated switching sequences for substations. This paper will present its application to instruction table generation for local load dispatching offices. In the case of dispatching office the huge number and individual particularity are the points at issue. The office has to deal with hundreds of substations and transmission lines, and make a few hundreds of switching instruction tables a month. Generating tables is one of the most difficult tasks because the uniqueness of substations such as structural differences and particular management rules have prevented us from completing knowledge base which will contain thousands of rules.
Our approach concerns the point for easy knowledge acquisition of the expert system, using an original learning algorithm named HCL (Hierarchical Concept based Learning). The HCL generates operational knowledge from cases in the past. This paper will discuss the fundamental problems of instruction table generation for dispatching offices and the solution to it by our approach. The method will be shown to construct and update its knowledge base automatically, using an actual network and operational cases of a dispatching office.