Research On Air Quality Prediction Technology Based On MISTS-LSTM

With the development of science and technology,the impact of air quality on people’s lives has become more and more serious,and the impact of PM2.5 is particularly prominent.Therefore,how to monitor PM2.5 more accurately and predict PM2.5 with high precision,so as to make a series of preventive measures in time,has become a very far-reaching research topic.However,the common traditional monitoring methods in today’s life are mostly single-point fixed methods,that is,using a certain point data to represent the overall PM2.5 concentration value in a large area.In fact,the data of one point is difficult to describe the distribution of PM2.5 in the whole area,and the established model often ignores the influence of various other characteristics in the time series processing of PM2.5 prediction,which leads to the prediction effect after entering the model prediction difference.Consequently,this thesis proposes a multi-sensor group mobile monitoring method based on laser sensors to obtain accurate data corresponding to the real location to cover the entire area,and for this purpose,a mobile PM2.5 acquisition hardware based on laser sensors was designed.In view of the large scale,high complexity and intractable characteristics of the collected data.This system uses the data divergence as the main distance parameter to perform cluster analysis to generate high spatial and temporal similarity regions.According to each partition data set,an attention mechanism is added to extract the primary and secondary features respectively to enhance the robustness of the model.Finally,the multi-layer LSTM(Long Short-Term Memory)algorithm is used to simulate the change rule of PM2.5 concentration in the global space sequence to establish the MISTS-LSTM(Spatiotemporal Similarity-Long Short-Term Memory)prediction model,and the prediction results are compared with LSTM,A comparative study of STS-BP(Spatiotemporal Similarity-Back Propagation)and STS-RNN(Spatiotemporal Similarity-Recurrent Neural Network)algorithms.The results show that the prediction model based on laser sensor and MISTS-LSTM can not only accurately predict the change of PM2.5 concentration value in the future,but also achieve accurate prediction in airspace.Compared with other algorithms,this model has the advantages of more accurate prediction results and higher applicability and wider effective coverage,it can better solve the situation where the predicted results of the site are inconsistent with the actual living environment concentration.