电力系统故障诊断神经网络专家系统的一种实现方式

对人工神经网络和专家系统结合应用于电力系统故障诊断问题进行了研究，提出了电力 系统故障诊断神经网络专家系统的一种实现方式。该结构方案中，采用三层前向BP网络作 为 故障诊断的核心部分，与传统的专家系统相结合组成混合式的神经网络专家系统。基于该方 案建造的故障诊断神经网络专家系统综合了专家系统和人工神经网络各自的优点，充分利用 专家系统的推理判断能力和人工神经网络的学习和容错能力，比单独利用专家系统或人工神 经网络的电力系统故障诊断系统具有更好的性能。

AN APPROACN TO BUILD EXPERT SYSTEMS WITN ARTIFICIAL NEURAL NETWORKS FOR POWER SYSTEM FAULT DIAGNOSIS

This paper describes how to integrate ANN with ES for power system fault diagnosis.A feasible approach is presented to build an ANNES for power system fault diagnosis. In theapproach.local mergence of ANN and ES is employed. Three layer BP networks are employed asmain roles to diagnose fault elements,an hybrid ANNES can be built by means of thecombination of ES and ANN. The advantages of ANN and ES are fully employed in the ANNES.Therefore,the ANNES is more powerful than a traditional expert system.