A PM2.5 measurement method based on image features and integrated neural network

Abstract

The invention relates to a soft sensing method for PM2.5 of air fine particles based on image features and an integrated neural network, which belongs to the field of both environmental engineering and detection technology. An atmospheric environmental system has many variables, nonlinear and complicated internal mechanism. Compared with single neural network, an ensemble neural network has betterability to deal with highly nonlinear and seriously uncertain system, and the real-time and high efficiency of PM2.5 prediction can be improved effectively by using image features as input variables.The invention aims at the problem that PM2.5 is difficult to predict with high precision and real-time. Firstly, the image features related to PM2.5 are extracted based on the feature extraction method. Secondly, the soft sensor model between PM2.5 and the image features is established by using the ensemble neural network based on the simple average method. Finally, the PM2.5 is predicted with the established soft sensor model and good results are obtained. The output results of the soft sensor model can provide timely and accurate information of atmospheric environment quality for environmental management decision-makers and the masses, which is conducive to strengthening the control of atmospheric pollution and preventing serious pollution.

Description

technical field [0001] The present invention relates to fine particulate matter PM in the air 2.5 soft-measurement method. Soft measurement is to select a group of input variables that are closely related to the main variable and easy to measure according to some optimal criterion, and realize the estimation of the main variable by constructing a certain mathematical model and relying on prior learning and memory. PM 2.5 The prediction of is of great significance in the prevention and control of air pollution, and the soft-sensing method is applied to PM 2.5 In the prediction of air pollution, it can not only save the cost of air pollution monitoring, but also provide timely PM 2.5 It is an important branch of the advanced manufacturing technology field, which belongs to both the field of environmental engineering and the field of detection technology. Background technique [0002] The Air Pollution Prevention and Control Action Plan issued in 2013 clearly stated that by...

AI Technical Summary

Problems solved by technology

Although the prediction accuracy of the artificial neural network is not high, the integration of the neural network will significantly improve the model accuracy

Most existing PMs 2.5 The prediction method cannot predict PM in real time because it is based on the data of the Meteorological Bureau that is not easy to obtain in real time 2.5 concentration

Images