Research On Intelligent Fault Diagnosis System Of Steam Turbine Unit Based On Data Stream

As the key equipment of the coal fire power station,the operation stability of the steam turbine generator set are directly related to the production cost,the power quality and the safety of the power plant.With the generator set in the direction of large scale and complex development,point object is increasing,resulting in a large,continuous real-time,stream flow data,which contains abundant information of equipment status.How to process the data to obtain useful information in intelligent fault diagnosis system of steam turbine generator set,is of great significance on maintaining the operational quality and the safety for the steam turbine generator set.Based on the above background and the project "Intelligent Diagnosis Subsystem of Generator",the research of intelligent fault diagnosis system of steam turbine based on data flow is carried out.The specific research contents of each part as follows:1.Based on the concept of data stream,the data model of data stream and the application of data stream are introduced.On this basis,a model of intelligent fault diagnosis system for steam turbine generator set based on data stream technology is established.At the same time,the function modules of the system are described,and the data stream forecasting model is analyzed.2.In order to solve the problem of steam turbine generator set anomaly detection,a mobile wavelet tree is constructed to realize the fast query of data stream.At the same time,to meet the requirements of real-time and accuracy in generator set anomaly detection,an anomaly detection algorithm based on prediction model,anomaly detection by constructing a sliding window with gauss regression algorithm model.The experimental results show that the processed algorithm can satisfy the requirements.3.Based on BP neural network,support vector machine and deep belief network,this paper combines four methods of feature extraction and selection,and studies fault features extracting of the steam turbine generator set.Sliding window is combined with uniform sampling to realize continuous prediction of data flow.The experimental results show that the data stream forecasting method based on prediction method and feature extraction method can meet the requirements of data flow in prediction accuracy and real-time processing,and can be used in the intelligent fault diagnosis system of steam turbine generator set for data flow prediction.4.Based on the above research,the intelligent fault diagnosis system of steam turbine is designed and developed.Through the design of the overall structure of the intelligent diagnosis system of generator set,the corresponding software operating environment is selected to realize functions,such as real-time condition monitoring and alarming,trend analysis,fault diagnosis and so on.Based on the actual project,this paper provides some experiences and thoughts on the development of fault diagnosis system and the research of data flow technology.