Research And Application Of AQI Online Monitoring Based On Neural Network

In the last few years,air environment quality has been a hot topic around the world.The current situation of air environment quality in China is not very good,and air pollution has brought a relatively large impact on people’s health.Air quality is generally assessed using the Air quality Index,which includes PM_2.5,PM₁₀,SO₂,NO₂,CO and O₃.Therefore,it is necessary to design a scientific,convenient and efficient air quality monitoring system.The project is divided into two parts:air quality prediction based on neural network and the development of air quality online monitoring platform.LSTM is used to build and train the model in air quality prediction.In order to improve the accuracy and fitting degree of the model in prediction,wavelet transform and LSTM neural network were proposed to fuse.Wavelet transform can effectively separate the low frequency information and high frequency information in air quality data timing sequence,and can deal with the fluctuation of data well.In the prediction of air quality data,the combination of wavelet transform and LSTM can give full play to the corresponding advantages,so as to make a better prediction of the trend of air quality.The online air quality monitoring platform USES technologies such as Spring Boot,Netty,Vue.js and Redis to realize data transmission,real-time air quality query,historical air quality query and timed statistics of site data.Through the combination of Redis cache database and TDengine data based on the Internet of things,the server side design using Netty realized the air quality real-time detection of multiple sites online.In order to evaluate the accuracy and fitting degree of the model,the root-mean-square error and R-square score were selected to evaluate the fitting degree of the model.After experimental verification and calculation,the RMSE error of the traditional LSTM model in predicting air quality data is 41.39,the RMSE error of R square is 0.9245,the RMSE error of the fusion model of wavelet transform and LSTM is 15.24,and the Score of R square is 0.9897,and the effect has been greatly improved.Figure 23;Table 13;Reference 54...