PM2.5 full coverage prediction method based on deep neural network

The invention discloses a PM2.5 full coverage prediction method based on a deep neural network, and the method comprises the steps: carrying out the preprocessing of air pollution concentration, meteorological data and land utilization data, which are obtained in advance, and dividing the data into training data and test data; a deep neural network prediction model STA-ConvLSTM is constructed, and the deep neural network prediction model STA-ConvLSTM is trained; the STA-ConvLSTM takes a CNN (Convolutional Neural Network) as a bottom layer, and spatial correlation of grid data is extracted through convolution; taking stacked multiple layers of STA-LSTM with a space-time memory unit and a space memory unit as a middle layer of the prediction model, and extracting features of time correlation and space correlation; and the last layer is decoded by using the CNN layer in combination with the features extracted by the STA-LSTM unit. The method has the advantages that multi-source heterogeneous data is fused, more