Research On Automatic Diagnosis And Early Warning System Of Hydropower Unit

The development and construction of hydroelectric power with the great development of the country,Large and medium-sized hydropower units have been put into production in large quantities,Increasing capacity of single machine,Because the hydroelectric unit is an important equipment for hydroelectric power generation,It is the main task of the power plant to ensure the safe and stable operation of the hydropower units.With the growth of the annual average generation growth,Grasp the health status of the hydroelectric unit at all times and find out the hidden danger and defect of the unit in time,and the maintenance of the hydroelectric unit is carried out.The economy can not only control overhaul,It can also prevent the shutdown due to failure of the unit of Serious economic losses,and It can also provide factual evidence for automatic intelligent diagnosis of hydropower units.This paper is to design a special early-warning system,event database and preprocessing data interface in the process of developing "rapid intelligent reaction system for running events of hydroelectric units" for the 5#unit of Huadian Wuxi River Power Plant of Zhejiang.It realizes automatic discrimination of gross error data,and provides facts and evidence for triggering intelligent fault diagnosis system to make fault diagnosis in advance.In this process,the main contents of this paper are as follows:(1)Summarize and analyze the typical faults corresponding to the analog alarm status.When the coarse system error data is found based on the actual data,the consistency,authenticity,and correctness of the data are discussed,and the preprocessing algorithm is used.Exclude gross abnormal data.This pretreatment algorithm is integrated into the development and design of data interface.After testing,it is found that the preprocessing algorithm can handle coarse system error data well.The preprocessing algorithm includes:over-range data processing and coarse system error data.(2)The SQL SERVER database is used as the background database of the intelligent system and the early warning system.The established database is the event database.The event database is designed in detail,including conceptual design,logical design,physical design of the event database and table partition design of the database.,to improve the performance of the event database and applications.(3)The three prediction models of ARMA model,grey model and BP neural network are realized by the algorithm,and the historical data of the warm water guide tile temperature are selected as sample data,and the prediction models are replaced respectively.The prediction error of ARMA and GM(1,1)model is smaller,and the BP neural network prediction model has a relatively large error..Through the average relative error of each model,it can be concluded that the grey prediction model has a better prediction effect on the upper guide and water guide temperature of hydropower units.(4)In comprehensive consideration,the GM(1,1)model group is applied to the research and development of the automatic diagnosis and early warning system for hydropower units.The real-time real-time early warning is carried out for real-time data,and the design of the middle and long term early warning algorithm for the historical data in the event database is designed.The early warning software is installed in the LAN of Wuxi river water power plant,the threshold is set by the threshold setting module.The early warning function of the early warning system is normal by using the method of changing threshold,and the correctness of the data is guaranteed by the trend prediction module.The system of early warning system linkage fault diagnosis system is real.The fault diagnosis is present.