Air quality prediction in weather forecast is a practical task of great significance.Using spatiotemporal sequence prediction in the field of machine learning to predict the future PM2.5 index is an important research problem.To solve this task,a two-layer LSTM is constructed to characterize the temporal dependencies within temporal air quality data and the spatial dependencies among air quality data from different monitoring stations.The two-layer LSTM takes the temporal air quality data from all monitoring stations in the past 12 hours as input,and outputs predicted PM2.5 for all monitoring stations in the next 6 hours.The experimental results show that the model achieves high classification precision,recall,and global accuracy on the test set,and indicate that the model can solve the PM2.5 prediction task well combined with the visualization analysis.
关键词
空气质量预测/时空序列/双层LSTM
Key words
air quality prediction/spatiotemporal sequence/Two-layer LSTM
维普
万方数据