The Design And Implementation Of Air Quality Index Prediction System Based On CNN-AGU

With the development of urbanization and industrialization,air pollution greatly influences urban residents’ health and daily life.The air quality index(AQI)is an international standard evaluation system of atmospheric environment quality and an important basis for measuring air quality.Therefore,constructing an efficient AQI prediction system to control air pollution effectively for the relevant departments has practical significance and social value.This paper proposes an AQI prediction model based on convolutional neural networks(CNN)and attention gate unit(AGU),CNN-AGU.For the time series characteristics of air quality and meteorological data,considering the temporal correlation of sequence data and the applicability of gated recurrent unit(GRU)for processing time series tasks,this paper proposes the AGU model based on the GRU model.AGU adds the attention mechanism(Attention)algorithm and data adjustment module(DAM)in the GRU structure,further enhancing the learning ability of the model.CNN has the characteristics of further mining data features.In this paper,CNN and AGU models are combined and applied to the AQI prediction task.This paper compares the CNN-AGU model with the other six models to verify the effectiveness of the CNN-AGU model,and the results show that the CNN-AGU model has minor deviation and higher prediction accuracy.This paper uses the Flask framework to build an AQI prediction system based on the construction of the CNN-AGU model.The system realizes the function of AQI prediction for the next hour,which provides an effective basis for people’s healthy travel and the scientific and reasonable decision-making by the environmental protection departments.Finally,through functional and non-functional tests of the AQI prediction system,the results show that the system can meet the functional requirements of users and has good availability and reliability.Therefore,the system has a certain application value.