Analysis Of Characteristics Of Power Station Boiler Drum Water Level And Research On Early Warning Methods

In recent years,the power industry is become the foundation of China's economic development.Thermal power generation plays a very important role in energy supply in China.With the core equipment of thermal power generation,the safety of utility boilers is very important."Drum water level”is one of the key indicators of boiler operation stage,the “lifeblood” of the boiler,and an important parameter to measure and determine the safe operation of the boiler.Drum is a large and complex system and there are many risk factors.Under the action of different factors,the water level has different response characteristics.At present,there is little research on the abnormal early warning method of drum water level,mainly the alarm technology after the accident happened.The abnormal drum water level will not only cause casualties,equipment damage and property losses,but also have a serious impact on life,production and even society.Therefore,it is necessary to study the risk analysis,characteristics analysis and early warning methods of drum water level.In this paper,through literature reading,field investigation,accident statistics,data acquisition,system dynamics,entropy weight-G1 method,random forest,BP neural network and support vector machine,etc.,and using VENSIM-DSS and WEKA platforms to study the water level characteristics and early warning methods of power plant boiler drum.Based on 2?100MW units of thermal power plant of Liaoning Fushun Petrochemical Company,the following research work is carried out:(1)Based on the work of accident statistics,mechanism analysis and risk factor analysis,the "person-machine-working-condition-management" factor is considered comprehensively.A risk assessment method of drum water level based on system dynamics is proposed.(2)Based on system dynamics,the SD model of drum water level risk is established,and the risk identification and risk assessment of drum water level are carried out by using VENSIM-DSS platform.The risk factors and risk development trend of drum water level are analyzed.The validity and sensitivity of the model are tested and analyzed to extract indicators for the water level characteristics analysis.(3)From the point of view of conservation of matter and energy,the mathematical model of drum water level mechanism is established,Based on VENSIM-DSS platform,a simulation model of drum water level is established,which combines mathematical model with system dynamics model.The characteristics of drum water level under typical factor disturbance are studied,the law of water level is analyzed and lay a foundation for the research of drum water level early warning methods.(4)The stationarity test and smoothing processing of the collected historical data of drum water level are carried out,and the drum water level prediction based on stochastic forest,BP neural network and support vector machine is proposed.The prediction results of the three methods are compared,and a RF prediction model with higher accuracy and better effect is selected,which lays a good foundation for drum water level early warning.(5)Based on the analysis of drum water level characteristics,the early warning index is determined.According to the actual operation of the boiler and the alarm degree of water level in the thermal power plant,the early warning grade of water level is divided.A practical early warning method for drum water level is put forward and applied.(6)According to different warning levels and types,early warning and pre-control measures are put forward to provide emergency and pre-control countermeasures for drum water level early warning,and puts forward some suggestions on the management and maintenance of drum water level system.