Research On A Neural-network-based Fault Diagnosis Expert System For Sluice

The sluice is the important component of water conservancy. If the sluice goes wrong, it will do a lot of damage to the water conservancy and the whole drainage basin. At the same time, it is difficult to find out the source of the sluice's fault and to repair it because the sluice is very complex. For these sakes it is a significant and essential task that the fault system of the sluice is developed .The research object of this paper is the sluice of the north brook of the Jiulong Rive. In this paper, I generally research on all kinds of methods of fault diagnosis and the trend of the technology of the fault diagnosis. At the same time I also research on the status of the north brook of the Jiulong Rive. On the base of these researches I give a neutral-network-based fault diagnosis expert system method and develop the fault diagnosis system in this method.This paper indicates that the standard BP algorithm's main weakness is the slow training speed. I present in this paper unite BP algorithm with variable step to solve this problem. This paper analyzes all kinds of integrated modes of neutral network and expert system. At last I propose serial connection mode . I diagnose the sluice's fault in the method. This paper saves information in database. This makes the building of the fault diagnosis system easy. The sluice has many fault sign and the fault type .In order to diagnose the sluice's fault, I adopt distributed diagnostic method in the process of the building of the fault diagnosis system and the system is made up of many neutral networks. This method makes the difficulty of building low and improves the speed of diagnosis. In the end, I program the software of the fault diagnosis system and simulate the system. The simulating result improves that this fault diagnosis method is feasible and ascendant.