A depth-learning air quality prediction method based on multi-model fusion

Abstract

The invention discloses a multi-model fusion depth learning air quality prediction method. Firstly, historical air quality data and meteorological data are obtained. 2, interpolating and normalizing the missing values of the historical air quality data; thirdly, a depth learning model based on seq2seq is constructed as a single-factor prediction model by using historical air quality data. Fourthly, using historical air quality data and meteorological data, a seq2seq depth learning model based on dual attention mechanism is constructed as a multi-factor prediction model. Fifthly, the single-factor prediction model prediction result of the air quality data, the multi-factor prediction model prediction result and the current meteorological data are integrated into the xgboost lifting tree forregression calculation to obtain the prediction value of the final air quality data, and the method of the invention can improve the prediction precision of the air quality data.

Description

technical field [0001] The invention belongs to the interdisciplinary field of computer science and environmental science, and in particular relates to a multi-model fusion deep learning air quality prediction method. Background technique [0002] Due to the progress of society and the rapid development of industry in recent years, a large number of serious environmental pollution problems have been caused, especially the air pollution has been aggravating. There are more and more inhalable particles in the air, and the air quality people are living in is gradually declining. Among the inhalable particles in the air, PM2.5 and PM10 are the most serious, which not only have a great impact on air quality, but also cause great harm to the human body. In order to reduce the harm of air pollutants to the human body, the analysis and prediction of air quality has important practical significance. [0003] Aiming at this problem of air quality prediction, many researchers have mad...

AI Technical Summary

Problems solved by technology

[0005] When monitoring stations acquire air quality data, there are usually many missing values ​​in air quality data due to problems such as equipment failures or network delays.

In traditional data preprocessing, especially missing value processing, most of the methods for filling missing values ​​are deletion, mean method, and proximity method. The accuracy of filling is poor. However, air quality data contains time series information, resulting The data is less accurate when training

Images