PM2.5 Concentration Prediction Based on Improved Manta Ray Foraging Algorithm Optimizing Support Vector Machine
SHI Weihua
LIU Qin
Abstract：PM2.5 has serious impact on air quality,whose prediction is affected by pollutants and the weather factors at the same time.And PM2.5 prediction is characterized by nonlinear and uncertain,the traditional prediction method has lower precision.Therefore,an air PM2.5 concentration prediction model based on improved manta rays foraging optimizing SVM is presented.First of all,a population initialization method based on Laplacian distribution is introduced to promote the population diversity.A weight factor adjustment based on Levy flight is used to improve the global optimization ability of the population.An individual disturbance mechanism based on the neighborhood center opposite-learning and Cauchy mutation is designed to avoid the local optimum.Then,improved manta rays foraging optimization algorithm IMRFO is to optimize support vector machine model,and an air quality predic-tion model IMRFO-SVM is constructed.Finally,The pollutant and meteorological data of a place are selected as experimental fea-ture factors,and the prediction factors are determined by Pearson correlation analysis for experimental analysis.The prediction re-sults of IMRFO-SVM is compared with that of traditional SVM,traditional MRFO optimizing SVM MRFO-SVM and traditional PSO optimizing neural network PSO-BP.Experimental results verify that IMRFO-SVM has higher forecasting precision and faster conver-gence rate.
Keywords：air quality predictionPM2.5 concentrationsupport vector machineLevy flightLaplacian distribution
Publication Date：2025-10-20
Online Publishing Date：2026-01-16(First online date of this platform, not the publication date of the document)
Pages：8( 35-42 )