Explanation-based generalisation method for learning network fault diagnosis rules

Artificial intelligence methods and knowledge-based approaches have been developed and applied to problems in power system such as planning, operation, control and fault diagnosis. In such systems, knowledge can be externally provided, or automatically acquired through the use of machine learning techniques. This paper reports an application of the explanation-based generalisation (EBG) learning method for the automatic generation of diagnosis rules for assessing faults in distribution ring networks. An important feature of EBG is that it produces generalised rules that are justifiable and correct. This paper also discusses the limitations of EBG with reference to the application example and possible methods of alleviating these problems