Application of chaotic prediction model based on wavelet transform on water quality prediction

Abstract

Abstract. Dissolved oxygen (DO) is closely related to water self-purification capacity. In order to better forecast its concentration, the chaotic prediction model, based on the wavelet transform, is proposed and applied to a certain monitoring section of the Mentougou area of the Haihe River Basin. The result is compared with the simple application of the chaotic prediction model. The study indicates that the new model aligns better with the real data and has a higher accuracy. Therefore, it will provide significant decision support for water protection and water environment treatment. IntroductionThe first paragraph after a heading is not indented (Bodytext style). Along with economic development and improvements in people's standard of living, bodies of water are facing more and more serious pollution and destruction. However, polluted water has certain self-purification abilities. It can reduce the concentration of pollutants to normal levels by physical, chemical and biological functions. Dissolved oxygen (DO) is the basis for studying water bodies' self-purification abilities. Consequently, research concerning trends of DO in water and predicting its concentration has substantial theoretical and realistic significance. Because of the complexity of the water environment and the limitation of single methods, combined forecasting methods have been widely used, such as the radial basis function (RBF) water quality prediction model based on particle swarm optimization, the BP neural network prediction model based on ant-colony optimization and the gray and neural network combination model [1][2][3]. These methods led to accurate predictions of water quality trends and concentrations of pollutants.Water environment systems are quite complex. The DO value could be affected by plenty of factors. As a signal processing technology, the wavelet transform has been widely used recently. The decomposition and reconstruction of original signals can lead to better extraction of the feature information of the original signal [4]. The chaotic prediction model based on the wavelet transform is established by combining the wavelet transform and the chaotic prediction model, which has simple calculations and high precision. Comparing it with the chaotic prediction model proves the validity of the new model.

A hybrid wavelet-Lyapunov exponent model for river water quality forecast

“…COD monitoring records are divided into two groups: calibration period or 'warm-up' period (total 250 data from the first week in 2009 to the 41st week in 2013) and verification period (total 30 data from the 42nd week in 2013 to the 19th week in 2014) (Grassberge & Procaccia 1983;Zhang et al 2016). Warm-up means the chaos feature of the low-frequency signal of decomposed surface water quality time series is identified at this period.…”