Research And Implementation Of Air Pollution Monitoring And Forecasting System Based On LSTM Neural Network

With the rapid development of China's science and technology and the continuous improvement of the economic level,the exploitation and use of energy in China continues to increase.Because of the lack of sophisticated energy use and sewage disposal methods,air pollution is becoming more and more serious.In order to understand the future air pollution situation in advance and formulate relevant countermeasures as soon as possible to contain the pollution in advance,it is necessary to establish a set of air pollution monitoring and forecasting system.In this paper,based on the time series characteristics of air pollution and meteorological monitoring data,considering the temporal correlation of sequence data and the applicability of LSTM neural network to processing time series problems,an air pollution prediction model based on LSTM neural network is studied.The model can memorize long-term historical calculation results and combine the current information to analyze the logical relationship between sequence data to achieve efficient air pollution prediction.Then,based on the air pollution prediction model,an air pollution monitoring and forecasting system is established to realize the six main air pollutants' air pollution concentration prediction,the future air quality assessment and the air pollution warning function,so as to provide effective basis for people's healthy travel and environmental protection departments to control the air pollution scientifically and reasonably.The main work of this paper has the following three aspects.First,understand the development process of air pollution prediction method,study the popular neural network prediction method under the current machine learning background,then study the long and short time memory neural network,analyze the characteristics of the network and the applicability to solve the time series problem.Second,the air pollution prediction model based on the long and short time memory neural network is designed and established.Based on the training data format of the model,the historical pollution monitoring data and meteorological monitoring data are processed preprocessed and the model input adaptively processed.The training data are input into the model for network training,and six mature air pollution prediction models are obtained.Third,the air pollution monitoring and forecasting system based on LSTM neural network is developed,and the mature air pollution prediction model is applied to the system The system collects air pollution monitoring data and meteorological data from the national air quality historical data website and meteorological website.Through the logical analysis of the system and the prediction analysis of LSTM neural network prediction model,the air pollution concentration forecast,the future air quality assessment and the pollution early warning function of the future period are realized.Finally,the function,performance and model prediction bias of the air pollution monitoring and forecasting system based on LSTM neural network are experimentally verified.The results show that the system can achieve the functional requirements of the system and meet the performance requirements of users.It has good usability and reliability.The prediction results of the LSTM air pollution prediction model are smaller than those of the real pollution concentration data,and the prediction accuracy is relatively high.Therefore,the whole system has certain application value.