Abstract: By experirients, at valves used at production lines in Shangbai Baosban lron and Steel Company some techniques of inspection on faults and acquisition of fault char- acteristics are obtained. The fault diagnostic expert system for servo valve is accomplished. The traditional symbolic reasoning is based on knowledge ripresentation in regular forms and pattertching technique. And artificial neural network (ANN) ,in which know- ledge is distributed over joints and reasoning is done by numerical calculation, combines representation and memory of knowledge with reasoning. Therefore, an expert system xintegrated fuzzy ANN with symbolic reasoning is constructed This expert system is open to knowledge acquisition and management. By experiments, it is proven that the expert system is maintainable and efficient