Fault diagnosis of complex devices has become an issue of very importance in modern industrial process automation. Many methods about fault diagnosis were presented in last 30 years, and intelligent diagnostic methods have been rapidly developed in recent years. Their applications, however, are limited due to their inherent defects in some fields. In this paper, a new intelligent diagnostic model which combines Artificial Neural Networks (ANN) with Expert System (ES) is developed. The architecture of its hardware and software are discussed. Its knowledge base, and knowledge acquisition are studied. The hybrid reasoning mechanism and layered diagnosing strategy are proposed. The diagnostic expert system based on the new model has been built for MK9-5 machines fault diagnosis. It has been shown that the new intelligent diagnostic expert system based on ANN has higher effect and liability than the common diagnostic expert system does. It is expected to be further developed and improved in order to apply it in practice.