Key Technology Research Of Atmosphere Particulate Matters Online Monitoring Instrument And Forecasting Model

With continuous economy growth in China, natural environment is under threat, especially atmosphere environment. Environment protection is always one of serious issues in public. Among the air pollution components, particulate matters attract more attention, for it is not only can influence visibility but also can really harmful to human health.So far, atmosphere particulate matters online monitoring still rely on import in China. Furthermore, the import machines usually can't satisfy accuracy requirement besides of high price and difficult maintenance, for the temperature and humidity are very different from north to south in China. This article has focuses on key technology research of particulate matters online monitoring instrument and forecasting model. The research mainly includes three parts.1) Development of atmosphere particulate matters online monitoring instrument: Based on ?-Ray technology, an atmospheric particulate matters online monitoring instrument was developed. ?-Ray is capable to measure density of particulate matters because it is easy to penetrate the medium obeyed by Beer-Lambert law. ?-Ray is safer than ?-Ray with lower radiation energy. The atmospheric particulate matters online monitoring instrument has four components: constant current sampling system??-Ray detect system?changing filter system and dynamic heat system.2) Accuracy analyzing of ?-Ray monitoring instrument: In the ?-Ray monitoring instrument, the key point of accuracy measurement is the mass-absorption coefficient. In the paper, we analyzed four factors that can influence mass-absorption coefficient which include physical structure?energy of ?-Ray?distance between ?-Ray source and filter?mass density of the filter.3) Model of particulate matters forecasting: In order to forecasting the density of atmospheric particulate matters, we proposed two non-linear neural network models, and analyzed thirteen factors that can influence particulate matters density. We built two artificial neural networks: BP(back propagation) and RBF(Radial Basis Function). We discussed the forecasting ability and error in both models. RBF trains faster than BP, and RBF have higher train precision than BP.