Research Of Data Processing And Fault Diagnosis Technology Of Medium-Speed Coal Pulverizer

With the change of energy structure in China,the demand for deep peak shaving of thermal power units is constantly increasing,which puts forward higher requirements for the stability and safety of unit operation.Medium-speed coal pulverizer is an important auxiliary equipment of boiler.The operation safety of coal pulverizer has an important impact on the normal operation of coal-fired units.However,there are still relatively few studies on the detection of operation status and fault diagnosis of coal pulverizer.With the development of computer technology,the combination of intelligent algorithm and industrial production is becoming more and more popular.Taking a coal pulverizer in a thermal power plant in Zhejiang Province as the research object,this paper establishes a fault diagnosis model for coal blocking of coal pulverizer based on signal analysis by analyzing and processing the historical operation data of coal pulverizer.Due to the bad points and errors in the production data,it is necessary to analyze and process the historical data before performing the big data analysis on the coal pulverizer.The indirect measurement method based on thermal balance method is proposed for the parameters with low accuracy such as primary air velocity at the exit of coal pulverizer,which greatly improves the accuracy of measurement.For the parameters of single or multiple measuring points,different accuracy optimization algorithms are used to improve the accuracy of the data,and the reliability of the algorithm is verified on the current parameters of the coal pulverizer.In order to improve the representativeness of the data,a method of identifying the stable working conditions using the moving relative standard deviation is proposed,which can effectively identify the stable working conditions of the equipment under variable working conditions.It is of great significance to improve the representativeness of the sample data.In order to extract the changing trend of the parameters,a noise removal method based on data autocorrelation is proposed.It can basically remove the periodic noise in the running data of the equipment and greatly improve the availability of the sample data.On the basis of data pretreatment,a fault diagnosis model for coal blocking of coal pulverizer based on signal analysis is established by analyzing the mechanism of coal blocking fault of coal pulverizer.Based on the analysis of fault mechanism and the results of feature parameter screening,more than 30 parameters related to coal pulverizer are extracted.Four characteristic parameters,such as temperature difference factor,pressure difference factor,valve opening factor and coal quantity factor,which have strong correlation with coal blocking fault,are constructed and normalized.A fault diagnosis model of coal blocking is preliminarily established.Then,the historical coal blocking data is used as training samples,and the weights of characteristic parameters are trained by the neural network optimization algorithm,and the final diagnosis model of coal blocking fault of coal pulverizer is obtained.The model is validated on the actual operation data of a coal pulverizer in a power plant.It can provide fault warning 3-8 minutes in advance for the operation and maintenance personnel of the power plant.It is enough for the operation and maintenance personnel to deal with the problem,avoiding the occurrence of the fault and verifying the validity of the model.