Study On PM 10 Concentration Prediction Of Cement Commercial Concrete Based On Data Mining

With the continuous improvement of national infrastructure,the construction demand of various large-scale civil engineering and urban buildings is increasing.Accordingly,the demand for building materials such as cement commercial concrete is gradually huge.At the same time,the pollution of atmospheric particles such as pneumoconiosis and haze is becoming more and more serious,which not only affects the production and life of operators in harsh environment,It also has a bad impact on the health of ordinary people and the development of cities.If the concentration of particulate matter in the workplace can be predicted timely and accurately,it can not only provide scientific planning for operators to enter and leave bad places,but also provide theoretical research for enterprises and relevant departments to scientifically prevent and control atmospheric particulate pollution.Based on this,this paper uses machine learning algorithm to mine the data set based on the PM 10 concentration and relevant data of a building materials enterprise in Tongling City.The specific process of the research is as follows:Firstly,this paper summarizes the research results and prediction research directions of air quality related factors at home and abroad,arranges and analyzes the application places of random forest algorithm,and expounds the relevant theoretical basis;Combined with the statistical theory,the monitoring data of PM 10 concentration in the workplace of Tongling building materials enterprises from December 2019 to June 2020 and the meteorological condition data of Tongling City in the same period are analyzed.The correlation between meteorological factors and PM 10 concentration is analyzed by using the correlation coefficient,and the causes of the impact are studied.The variation law of PM 10 concentration in building materials company over multiple time spans such as month,day and hour is analyzed;Secondly,complete the preprocessing of the original data,fill,clean and merge various original data,then screen the characteristics of the relevant factors affecting PM 10 concentration through the correlation coefficient method and characteristic interference method,and cluster the meteorological characteristic data by using the cluster analysis algorithm,so as to prepare for establishing different prediction models for different cluster clusters in the future;Then,the decision tree in the random forest algorithm is weighted and the parameters are optimized,and then the PM 10 concentration prediction model based on the improved random forest algorithm is constructed.The four indexes of R2,MAE,RMSE and MAPE are used as the criteria to evaluate the prediction accuracy and error of the model;Finally,combined with the historical monitoring of PM 10 concentration and meteorological data,the prediction results of the improved random forest model and other prediction models on the same data set are compared,so as to objectively evaluate the prediction performance of the established model.The results show that the PM 10 concentration prediction model based on the improved random forest algorithm proposed in this paper has higher prediction accuracy.The prediction result of this model has the largest determination coefficient(R2)on the same data set than linear regression model,decision tree regression model,SVM regression model,KNN regression model and BP neural network,which is 0.8730.Compared with the unmodified random forest model,the decision coefficient increased by 0.0362,and RMSE and MAPE decreased.Therefore,the model has high accuracy and excellent prediction performance,and can be used to predict PM 10 concentration.Figure 27 table 15 reference 71...