Study On Fault Diagnosis Technology For Floodgates Maintenance Automation Of Hydro Project

An industrial automation system mainly includes three technical domains, namely, control, maintenance and technical management. The three domains are interconnected and interacted. In order to meet the comprehensive needs of performance, reliability and benefits, they should be integrated and studied as a whole, and the Intelligent Control-Maintenance-Management System (ICMMS) has emerged. In these three domains, the maintenance automation level is far below than the other two. Nowadays, equipments and systems are becoming more and more complicated, reasonable maintenance strategies and methods are very important in ensuring the safety of equipments and reduce the cost, and then improve the company's competitive ability. After analyzing and summarizing the research results of previous works, the focus has been put on the maintenance domain of ICMMS. Combining with neural network, fuzzy logic and expert system, the methods to analyze, design and implement the maintenance automation have been thoroughly discussed in the framework of ICMMS. A novel idea has been presented that the fuzzy expert system, fuzzy logic system and neural network are integrated and applied to the fault diagnosis and maintenance automation of floodgate integrated automation system of Geheyan hydro power station. The structure,analysis and design methods and steps of the maintenance domain in the framework of ICMMS are all detailed discussed. As an example of analysis and design methods of maintenance automation of ICMMS, the reference function models of Geheyan integrated automation system are presented. The failure mode and effects analysis (FMEA) of system equipments are also provided.The reliable operation of floodgate integrated automation system depends on the acquired data. For some reasons, parameter failure often happens and it is an impending problem to deal with in industrial automation system. The conventional method is to replace the failure parameter with the rated value. However, this method cannot work very well when the system in a different operation state. RBF Neural Network has been proposed in this thesis to deal with parameter failure in ICMMS. The field tests have demonstrated that the trained neural network can perfectly estimate the failure parameter, and it is practical in the real time system. It also provides a novel and effective method to deal with parameter failure in ICMMS.The fault diagnosis strategies of Geheyan floodgates integrated automation system are presented. Object oriented knowledge representation method is put forward to deal with the multi-hierarchy fault diagnosis structure of floodgate integrated automation system. Furthermore, fuzzy expert system is employed to representing the uncertainty knowledge of floodgate integrated automation system. After discussing the advantages and disadvantages of fuzzy logic system and neural network, the thesis tries to integrate fuzzy logic system and neural network in the fault diagnosis of floodgate integrated automation system. The simulation and application results both demonstrated the novelty and feasibility of this idea.The theory research results are applied to the floodgates integrated automation of Geheyan hydro power station. It is introduced that the system structure and functions of floodgate integrated automation of Geheyan hydro power station. The structure of system database is also discussed for the maintenance automation of floodgate integrated automation of Geheyan hydro power station, and the system detailed design is accomplished. The field test and application results are also provided.