Study On Condition Monitoring And Fault Diagnosis For Coal Pulverizer Of Power Plant

The thermal power based on coal-fired will still be the main power for meeting the electricity demand in China. Along with the development of power industry, increasing the competitive power of power plant becomes thermal power trade's aim. Enhancement of the operation stability of thermoelectric generating set, development of the condition monitoring and fault diagnose technology are the requirement of reform of power trade.Coal pulverizer is significant as an auxiliary equipment of boiler. The methods of condition monitoring and fault diagnose are studied in this thesis. First, a method of neural network prediction model which based on a stable and fast Levenberg-Marquardt algorithm for real-time date of thermal power plants is presented. Three different models are established to research based on this method, and three kinds of prediction models for one parameter which according to operating characteristics of different parameters are used to predict the trends of parameters. Second, a new method of estate evaluation which contains the results of sequential prediction based on neural network is presented. It makes the estate evaluation more comprehensive and earlier. Third, a mode of fault diagnose using expert system is established and the repository of expert system for coal pulverizer is designed.Furthermore, a system of condition monitoring and fault diagnose for Coal pulverizer is developed in this thesis. The system can monitors the coal pulverizer A of generate set 1 on-line, and provides a convenient way to have the messages of the coal pulverizer's function for user. Besides, it has the off-line diagnostics function, and provides an easier way for user's study. In conclusion, it is important for the development of condition-based maintenance.