In recent years,PM2.5,arch-criminal of fog weather,has aroused extensive attention.A variety of toxic viruses which is adsorbed by PM2.5 in the air can be directly inhaled into the respiratory system and directly into the blood,that will increase the disease of lung cancer and heart disease.Therefore,it is very necessary to analyse PM2.5 timely.First of all,based on the data of Xi'an's air quality index from 2013.3.1 to 2016.2.29,we analyse the temporal and spatial distribution of Xi'an.We can draw the conclusion that PM2.5 is the highest in winter and the lowest in summer,and it is gradually improved this several years,but the content of PM2.5 is still not optimistic at present.Secondly,this paper further studies the relationships among PM2.5,AQI,meteorological factors and heating time with correlation analysis and ridge regression.We can draw the conclusion that SO2,NO2,and CO are the important gaseous objects to form the PM2.5,and SO2,NO2,PM10,CO,dew point,visibility and heating time are more significantly in the all influencing factors..Finally,the paper forecasts PM2.5 per hour with the method of neural network;wavelet analysis with neural network;genetic algorithm with neural network;and genetic algorithm,wavelet analysis with neural network analysis;respectively.And those methods' effect and accuracy is compared with error of test the network respectively.We can draw the conclusion that the effect of genetic algorithm combined with wavelet analysis and neural network is better than neural network using by other authors. |