Method for predicting multiple atmospheric pollutants in multiple cities based on four-dimensional directed GCN-LSTM model

• Main Authors: ZHANG HUIYAN, CUI XIAOYU, WANG LI, XU JIPING, BAI YUTING, TANG QIANHUI, YU JIABIN, WANG XIAOYI, ZHAO ZHIYAO, WANG ZHAOYANG, SUN QIAN
• Format: Patent
• Language: Chinese; English
• Published: 23.06.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FORADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORYOR FORECASTING PURPOSES; ELECTRIC DIGITAL DATA PROCESSING; INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIRCHEMICAL OR PHYSICAL PROPERTIES; MEASURING; PHYSICS; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE,COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTINGPURPOSES, NOT OTHERWISE PROVIDED FOR; TESTING

The invention relates to a method for predicting multiple atmospheric pollutants in multiple cities based on a four-dimensional directed GCN-LSTM model, and solves the problem that the concentration of multiple atmospheric pollutants cannot be accurately predicted at present. The method comprises the following steps: carrying out correlation analysis on concentration data of multiple atmospheric pollutants in multiple cities, and determining that the data have correlation; establishing a graph of a four-dimensional directed GCN model according to the geographic position of the city; performing spectral decomposition and tensor operation of the four-dimensional directed GCN model to obtain a graph Fourier coefficient and a graph Fourier basis; improving a graph filter of the four-dimensional directed GCN model; and an LSTM network architecture is introduced, and a four-dimensional directed GCN-LSTM model is constructed for atmospheric pollution prediction. According to the method, a four-dimensional directed g