Font Size: a A A

Air Quality Assessment And Prediction Based On Machine Learning

Posted on:2020-10-14Degree:MasterType:Thesis
Country:ChinaCandidate:S GaoFull Text:PDF
GTID:2381330572999256Subject:Mathematics
Abstract/Summary:PDF Full Text Request
With the acceleration of economic construction and urbanization,people's quality of life is increasing day by day,but the air pollution caused by the massive consumption of energy and pollutants is becoming more and more serious,and the haze weather is increasing day by day,and it is a daily life and physical and mental health for human beings.A serious threat has arisen.The Air Quality Index(AQI)is an important basis for measuring air quality,so air quality prediction is extremely important for guiding people's lives and work.Air quality data is a non-linear time series whose high complexity makes it difficult for traditional methods to achieve the desired results.In recent years,the rise of intelligent optimization algorithms has provided an idea for solving such problems.Its strong search ability,easy to find the characteristics of the optimal solution,has been favored by research scholars.In this paper,we use the support vector machine in machine learning,group optimization algorithm and neural network to study the air pollutant monitoring data between2015 and 2018 in Taiyuan City.The main research contents are as follows:(1)The K-nearest neighbor algorithm(KNN)is used to classify and predict the air quality levels of Taiyuan City and Datong City.In the process of prediction,the air quality is classified into two categories and multi-classification with 70% as the threshold point,and the error analysis of the prediction results.The experimental results show that the classification prediction is close to the real value.Further analysis of the prediction results,the establishment of the Pearson correlation coefficient matrix,the correlation between the daily mean values of various pollutant concentrations was obtained,and the Pearson correlation coefficient between the AQI values and the various pollutant concentrations was found.The main pollutants affecting the air quality in Taiyuan City,thus controlling air pollution from the source.(2)Using the hybrid model of thought evolution algorithm(MEA),particle swarmoptimization(PSO)and genetic algorithm(GA)to optimize BP neural network parameters and predict the future air quality index of Taiyuan City.The experimental results show that the MEA-PSO-GA-BP algorithm has better search speed in terms of prediction accuracy,error rate and reliability,and has good validity and feasibility.It has certain practical significance for predicting AQI.(3)The moth-fire-fighting optimization(MFO)algorithm is easy to fall into the local optimum and the support vector machine(SVM)penalty function c and the kernel function parameter g are difficult to find.Based on the advantages of support vector machine(SVM)and the need for global optimization,a combination of SVM and MFO(MFO-SVM)is proposed.The daily air quality index(AQI)of Taiyuan City and Datong City of Shanxi Province was selected to verify the feasibility and effectiveness of the algorithm.The experimental results show that the relative error of the MFO-SVM algorithm is close to zero,and the predicted value is closer to the actual value,which can effectively predict the air quality index.Two prediction models proposed in this paper: MEA-PSO-GA-BP and MFO-SVM for air quality prediction.It provides a new idea for air quality prediction,and provides a scientific and rational theoretical basis for the prevention and control of air pollution and effective measures.
Keywords/Search Tags:K-Nearest Neighbor algorithm(KNN), BP neural network, Moth-flame optimization(MFO), Support Vector Machines(SVM), air quality prediction
PDF Full Text Request
Related items