.600 Mw Steam Turbine Intelligent Fault Diagnosis System

The steam turbine is one of the main equipments in the thermal power plant. With the development of national economy, various of high parameter, large capacity steam turbine-generator sets are putting into production constantly. This not only enhances the power productivity but also reduces the cost of labour. But once the fault of the sets happened, the economic losses will be multiplied, and social influence is immeasurable. So, it is particularly important to develop advanced state monitoring and diagnosis system. In this paper, the fault intelligent diagnosis system for 600MW steam turbine is researched and developed with LabVIEW virtual instrument development platform.This system is used in 600MW steam turbines. The functions of monitoring, displaying, alarm, data storage, automatic acquisition for fault symptoms and fault diagnosis for steam turbine working state are realized. Database system for steam turbine state monitoring is developed with Access 2003 as systemic database in this paper, and the realization of LabVIEW access database is through LABSQL toolkit. In fault diagnosis,all the fault symptoms are acquired automatically, thus the system can achieve the timeliness and accuracy. The main fault symptoms include the components and characteristics of frequency, shape and moving direction of axis orbit, the variation of vibration amplitude and phase, the relationship between vibration and sets working conditions such as exciting current, active power, speed, etc.There are various expression methods of knowledge. After the analysis of advantages and disadvantages of the knowledge representation methods of production and framework, framework as knowledge representation method and production as reasoning method are adopted. The inference engine is realized by selecting structure and sequence structure. In addition,Techniques of intelligent fault diagnosis based on neural networks is also discussed, which including the neural network structure design, network training and the method of realization in LabVIEW, and the neural network fault diagnosis is assistant diagnosis system. Finally, the method of knowledge representation and the system have been verified and indicated through computer-simulated data that they had highly practical value.