Research On State Evaluation And Prediction Of Power Plant Equipment Based On Real-time Reliability Analysis

As power plant unit becomes larger capacity, higher parameters and higher automation, systems of power plant unit become more complex. The factors that can affect unit operating are harder to predict than ever. And these make the requirement of reliable operation of generating units become much higher. At the same time, the smart grid development and the energy conservation and emission reduction both need power plant unit operating at the high safe, reliable and efficient way. Thus, the research for real-time reliability analysis of power plant equipment is badly needed, in order to guide the operating and maintenance work.Traditional reliability analysis method is based on the performance declining trend of the equipment through the whole life cycle, and its result is the probability of the equipment to perform its required functions for a given period of time. The result can not meet the need of the real-time status evaluation of equipment. For the actual demand from the power plant, the relationship between the monitoring parameters and the equipment reliability is researched, and the real-time reliability evaluation model based on deviation of operating parameters is established, in which the equipment real-time reliability level is quantify by the decrease of the operating reliability. The principal component analysis model is used to eliminate the correlation between multiple parameters. The unit system is divided into components level, equipment level and system level to evaluate the reliability level by level.At the same time, the trend prediction model of the real-time reliability is also established, using the secondary exponential smoothing method to predict changes in operating parameters. By predicting the short-term trend of the real-time reliability of the unit, the deterioration of equipment can be detected in early stage, and measures can be taken to avoid or reduce accidents.Through using in the actual case, the application proves the method to be effective and feasible in the state evaluation and prediction for power plant equipment.