Centrifugal Oil Delivery Pump Fault Diagnosis System

This paper firstly discusses the necessary of fault diagnosis and the meaning, resource and goal of fault diagnosis expert system for pumps, and reviews research situation home and abroad in the field of technical diagnosis, maintenance system. Then it introduces briefly basic knowledge on expert system, including definition, basic features, normal structure, classification, and so on .In this paper ,the application of neural network fault diagnosis expert system to centrifugal oil delivery pumps is studied concretely based on a discuss deep on the feature of centrifugal oil delivery pumps; and this paper puts forward a new idea of building combined neural network ,which is composed of basic neural networks just like Back Propagation neural network and Competitive neural network based on self-organization and self-adaptation. The basic neural networks are linked with each other in the way of layers ,and each layer corresponds to its own fault property vector set (or fault omen vector set) -part of the whole fault omen space that is constituted of fault omen vector sets . Based on this idea, this paper is trying to integrate BP network and competitive network into one neural network as the learning and reasoning core of fault diagnosis expert system for centrifugal oil delivery pumps and built up its initial frame. Based on the neural network structure in the paper, the object-oriented knowledge representation thought about neural network is deeply discussed and this paper trys to realize it with the neural network structure. To some extent, it is realized, but not mature, this paper just puts forward an idea, and gives some sugestions . At last, the paper prospects the coming researching work.