Remote Fault Diagnosis Expert-system For Loader Based On ANN And FUZZY

Researching and developing the information-based loader with the independent intelligent property right have an important significance on promoting the high-tech content, the market competition ability, the fast-speed development of the loader and the engineering machinery in our country. Aiming at the trait of this project, with using the theory of artificial neural net (ANN) and fuzzy system, this dissertation has carried on further investigation on loader's diagnostic algorithm and offered the global structure and principle of the loader remote service and diagnosis system. Basing on the theory of artificial neural net (ANN) and fuzzy system, this dissertation has deducted a hierarchy-evaluating model of remote loader diagnostic expert system. Combined the fuzzy system with the artificial neural network (ANN), the approach of establishing fault-diagnosis model for loader system was put forward. Established a fuzzy-system that can update membership functions by itself and proposed a new parallel sub-ANN approach which makes the training rapid and accurate. This model has made the calculation workload and size of the knowledge-repository reduced, and it has a strong ability for the system denoising, it can refine the feature that is only sensitive to the fault from fault-system. Simulated experiment show that this kind of model has good recognition effect and low false ratio. The dissertation describes the key function model and introduces the function modules of the software and several ways to realize the diagnosis system according to the software-realization of loader remote fault diagnosis system, then offers the detailed explanation to the key technology among them. With applying this remote service and diagnosis system on LG's ZL50G loader, this system has achieved some success. Finally, this dissertation introduces the summary and the prospect of successive work of the project.