城市大气污染物浓度预测方法研究

城市大气污染物浓度预测方法主要有2类,即以污染物排放为基础的模型和以与污染相关的环境因素为基础的模型.本文分析了后者中的3种预测模型,并利用BP神经网络模型对城市的SO2浓度进行预测,对不同预测模型的结果进行比较分析.

Approach to more efficient forecasting of concentration of air pollutants in urban areas

The present paper is mainly engaged in the study of the three efficient models for forecasting the concentration of air pollutants based on the related environmental factors. The models for forecasting are mainly based on two aspects. One is to predict the air pollution emission; the other is to analyze the related environmental factors. In the BP model, four factors are involved, including coal consumption in industries, population density, traffic volume and trattorias distribution, which are generally taken as the model inputs with the concentrations of SO_2 taken as output. In recent years programs on learning and training the BP network have been introduced through the environmental program-held training network and gradually turned into computer and internet- offering training programs. With new data and information increased, the forecasting results and achievements have been gained. The calculated results of BP model are 0.0123, 0.0281, 0.0797, 0.0919 and 0.0911 respectively, whose relative errors are between -56.07% and 17.82%, whereas that of the other models are between -35.71% and 35.71%. The differences among the various models are also compared in the paper. BP model is easily to use with the development of computer technology. But the common problems of the models mentioned in this paper are that the related errors are not so satisfying. So it is urgent to set up a continuous and long-term monitoring system so as to predict the concentration of the air pollutants in the urban areas.