Environmental data forecast and expression method based on wavelet neural network

This paper proposed a new forecast and expression method for environmental data. It's differ from other data types expression and forecast; Environmental data has very complicated influence factors. The change of phase is inconspicuous. It's difficult to extract valuable information from time and frequency domain. Usually, Environmental data can be expressed by of average value, maximal value and minimal value of phase. This paper proposed a novel K-Line expression method based on wavelet transformation and data compression for Environmental data forecast. The data saved has clear data structure, physical meaning and the nature information of environment. In addition, this paper proposed a new method for Environmental data forecast based on wavelet neural network. The general methods (such as BP and RBF) have a big forecast error and low convergence speed because of complicated factors. This paper adopts characteristic wavelet base and WNN. COD data of two year has been processed for forecast and evaluation. The experiment indicates that the forecast result has been improved remarkably for history data testing as well as practice application in all seasons. It provides scientific basis for the environment monitoring and management department.