An improvement of PM2.5 concentration prediction using optimised deep LSTM

Air pollution poses a serious threat to human health and the environment worldwide, of which particulate matter (PM2.5), receives an increasing attention with deeper recognition of human health risk. In this paper, we proposed a method for optimising the deep long short term memory (LSTM) model to improve the quality of PM2.5 concentration prediction and used it for PM2.5 concentration prediction. The parameters of the optimised deep LSTM model were determined by using the genetic algorithm, and were applied to predict PM2.5 concentration, thus achieving better results than when the genetic algorithm was not used. The predicted PM2.5 concentration results of the optimised deep LSTM model were compared with the recurrent neural network (RNN) and gated recurrent unit (GRU) models, respectively, showing that the LSTM model had improved performance. This method would possibly contribute to enrich noble solutions in the aspect of air-pollution prediction.