Based on the analysis of main problem in existing methods for urban air pollution forecasting,a new idea based on Bayesian Network was proposed,with preliminary discussion of technical route as well as key challenges in putting the new idea into practice.

With the development of society and economy and the population expansion of the city, air pollution becomes more and more serious, therefore, it is significant to study air pollution forecasting so as to understand environmental variation trend, prevent and control air pollution.

An SVM regression (SVMR) model for atmospheric pollution prediction is developed according to seven forecast factors, including the daily average pollutant concentration of previous day,daily average wind speed of previous day,etc.

According to the deficiency of traditional BP model in the slow training speed, the local minimum and the uncertainty on the best structure of neural network, BP model is improved and applied to urban atmospheric pollution prediction.

A design idea on establishing a kind of applicable air pollution forecast system is proposed according to the existing bases and conditions and objectives.

Based on the Regional Atmospheric Modeling System and by means of Grid Nesting technique, a nested numerical forecasting model system, the City-scale Meteorology Model(C-βM) nested to the Regional-scale Meteorology Model(R-αM), has been developed, which can supply meteorology data for air pollution forecast.

By the analysis of air pollutants (SO 2,NOx,TSP) data measured by the air pollution monitoring system in last three years and the meteorological data collected by Qingdao Meteorological Station in the corresponding period, the method of air pollution forecast was investigated.

The basic principle of BP neural network is briefly introduced, and the applications of BP neural network to air pollution prediction in China are reviewed in detail.

And the air pollution prediction is just this class of problem.