A Method Of Air Quality Prediction Model

It is known to all that air pollution is harmful to health.Recently,people's desire to predict air quality remain unchanged.With the establishment of a great number of air quality monitoring system,air quality data and relevant meteorological data have become abundant and this made air quality prediction become possible.Under such background,a new air quality prediction model is proposed.To design this model,21 feature dimensions are considered.In order to improve the efficiency of this model,the principal component analysis(PCA)is used to reduce the dimension of the data.Although,the core part of the model is still based on Back propagation neural network(BP neural network),an improved Levenberg-Marquardt(L-M algorithm)algorithm based on Bayesian regularization is designed,Genetic algorithm(GA)is also used to optimize the parameters of BP neural network.Besides,double hidden layers topology is created.All these three changes improve the defects of BP neural network.At the end of this paper,details of the experiment are expounded and the experimental results prove the feasibility of the proposed model.Compared to air quality prediction model which based on the traditional BP neural network,the proposed model has a significant improvement in convergence speed,training accuracy and prediction ability.