Forecast Of First Wind Volume Of Coal Grinder In Thermal Power Plant Based On Hierarchical Clustering Algorithm

At present,the economy and safety of fossil-fuel power station boilers depend on the stability of combustion.In the boiler combustion system of the thermal power plant,the primary air system of the coal pulverizer system is an important system,to ensure the transportation of pulverized coal and the combustion of the boiler.The primary air volume is the most important parameter of the primary air system.The arrangement of air supply is very important to the utility of boiler combustion.Therefore,it is necessary to reasonably predict the primary air flow of the coal pulverizer,to ensure the controllability of the primary air flow,and improve the accuracy of prediction,estimation of the primary air flow of the coal pulverizer.It provides data support for on-line monitoring and optimizing operation of coal pulverizer primary air volume.In order to predict the primary air volume of coal pulverizer in the thermal power plant,this paper selects 10 operating parameters,which are related to the primary air volume of the coal pulverizer,and then processes the collected data by means of grey relational grade analysis.It can select and determine 5 factors,which have a greater correlation with the primary air volume of the coal pulverizer,and then use the hierarchical cluster analysis algorithm to classify the historical data of 5 factors,which affect the primary air volume of the coal pulverizer.Finally,according to the clustering result,the historical data,which is similar to the forecast day condition,is selected as the training sample,and the test sample is the primary air volume value of the forecast day.BP neural network simulation model is established by Matlab software,to predict the air volume once,and to test the error of the predicted value.The innovation of this paper is to predict the primary air volume of coal pulverizer in the thermal power plant,and use K-MEANS algorithm combined with AGNES Algorithm to do hierarchical clustering analysis,which can make clustering accurate and fast.The results show that the relative error of the prediction value can be controlled within 10%,and the prediction value can meet the actual operation requirements of the coal pulverizer in the thermal power plant.