A National Urban Air Quality Prediction Method Based on Group-Aware Graph Neural Network

Abstract

The invention discloses a national urban air quality prediction method based on a group perception map neural network, including: 1) preprocessing urban air quality monitoring data and weather data, and using a sliding time window to obtain a training data set; 2) according to The geographical distribution of cities constructs a city map and models the spatial dependencies between cities; 3) uses a differentiable grouping network in the encoder to map all cities to a fixed number of city groups, and uses the group relationship encoding network to construct a city group map, Modeling the correlation between city groups; 4) Using the message passing mechanism on the city map and city group map to model the dependencies between cities and city groups respectively; 5) After the decoder receives the encoded output, it realizes end-to-end predict output. The present invention combines group perception map neural network and codec architecture to predict national urban air quality, and has broad application prospects in the fields of health, urban planning and the like.

Description

technical field [0001] The invention belongs to the field of urban air quality forecasting, and in particular relates to a national urban air quality forecasting method based on a group perception map neural network. Background technique [0002] Due to the rapid advancement of industrialization and the increase in pollution emissions from motor vehicles, air pollution incidents have occurred frequently in many places in recent years. Solving the problem of air pollution can not only protect public health, greatly reduce the incidence of respiratory and lung diseases, but also show a good city appearance to the outside world and enhance my country's international image. The government has established air quality monitoring stations to monitor local air quality in real time, calculate the air quality index (AQI) through standard formulas to quantitatively describe the local air quality level, and release relevant information to the public simultaneously. The researchers beli...

AI Technical Summary

Problems solved by technology

However, the existing deep learning-based urban air quality prediction methods only consider the spatial dependencies between cities that are closer, and do not consider the implicit dependencies between cities that are far away (for example, affected by factors such as terrain, some geographically separated Cities farther away may have similar air quality)

Images