Research on PM2.5 Forecasting Methods Based on Deep Learning——A Case Study of Suzhou City
The study investigates the relationship between air quality and meteorological factors in Suzhou City.PM2.5 concentration in Suzhou was efficiently and accurately predicted based on different deep learning models,which provides a scientific basis for air pollution control and risk avoidance.Utilizing historical air quality and meteorological monitoring data from 2014 to 2022,the study employs three deep learning neural network models:RNN,LSTM,and GRU.It compares the performance of these models in predicting PM2.5 concentrations at differ-ent time steps and analyzes the importance of various feature factors.The results indicate that the predictive per-formance of the models,ranked from highest to lowest at different time steps,is as follows:GRU>LSTM>RNN.The GRU model exhibits the best predictive performance at a time step of 40,achieving average absolute error,root mean square error,and normalized root mean square error of 8.68,11.50,and 0.10,respectively.Features highly correlated with PM2.5 include PM10,air quality level,and air quality index,which are identified as important features of the prediction model.Although wind power and weather characteristics are less correlated with PM2.5,they can significantly improve model performance,so they are also important features.The highest temperature,the lowest temperature and the wind direction are not important characteristics.
NETL
NSTL
万方数据
