Prediction method for PM2.5 concentration of fine particulate matters in air based on a stack selective integrated learning device

Abstract

The invention relates to a prediction method for PM2.5 concentration of fine particulate matters in air based on a stack selective integrated learning device. The concentration of six pollutants in the air per hour in the past 24 hours and six meteorological indexes totally account for 24 groups and 12 characteristics to serve as input, and a PM2.5 concentration predicted value is obtained. The model is realized in a three-stage framework and comprises the following steps of: firstly, creating a plurality of base learners by properly selecting environmental factors, time factors and training samples; secondly, deleting the negative basis learners in the three categories according to a dynamic threshold value by adopting a trimming technology; and finally, integrating the selected forward basis learners by adopting a stacking technology so as to predict the future PM2.5 concentration. Compared with an existing method, the method has the advantages that prediction errors and the difficulty degree of data sources are obviously improved, people can be guided to travel healthily, and the government can be assisted to limit automobile flow, exhaust emission and the like.

Description

technical field [0001] The present invention utilizes a stacked selective integrated learner to establish fine particulate matter PM in the air 2.5 Concentration prediction model, by taking the concentration of 6 pollutants and 6 meteorological indicators in the air per hour in the past 24 hours, a total of 24 groups of 12 characteristics are used as input, and the PM in the future time 2.5 Concentration is predicted. via PM for future moments 2.5 The accurate prediction of the concentration can not only guide people to travel healthily, but also assist the government to limit the flow of cars, exhaust emissions, etc. Airborne fine particulate matter PM based on stacked selective ensemble learner 2.5 Concentration prediction methods belong to both the field of air environmental protection and the field of machine learning. Background technique [0002] PM 2.5 Refers to particulate matter with a diameter less than or equal to 2.5 microns floating in the atmosphere. Becau...

AI Technical Summary

Problems solved by technology

Prediction by this method is significantly improved in accuracy compared with existing methods, and it solves the problem of PM 2.5 The concentration is difficult to predict and control, which can provide reference for government decision-making and mass travel;

Images