Research And Application Of Uncertainly Coupled Model Of Water Quality

Water environment is an intricate and huge system filled with uncertainties. It is great significant to research and apply the uncertain model of water quality for expressing the various features in the system more exactly and substantially. But, the research and apply have often confined to one-dimensional and steadily uncertain model of water quality which supposed the river way is regular for a long time. Since the deterministic model of water quality, Monte Carlo stochastic method and artificial neural network (ANN) have complementary advantages, an uncertainly coupled model is created by the dissertation which integrates these means in itself and the coupled model is in point to natural stream flow. The integration has come true in MATLAB. The applying study of Baiyin section of Yellow River and the comparative results with other models demonstrated that the coupled model studied in the paper can simulate complex river water quality and the imitation results can furnish more information.1 The main conclusions and findings are as follows:(1) The deterministically mathematic model of water quality, Monte Carlo stochastic method and artificial neural network are integrated in a model to create the uncertainly coupled model of water quality which enriches the study methodologies of uncertainly model of water quality and contributed substantially to the understanding and mastering the change of complex water environment; at the same time, the defining of uncertain behavior of other similarly intricate environment system can also learn the study thought and means.(2) The deterministic model of water quality, Monte Carlo stochastic method and artificial neural network are integrated in MATLAB by the coupling of ANSYS innovated and MATLAB in the paper. The coupling model is used to imitate the main pollutants transport and fate in Baiyin section of Yellow River. The results indicate the average simulation error is 9.6% against COD_Cr, NH₃-N and Volatile Phenol. The model is so accurate that it can be utilized to forecast the water quality of natural stream flow. (3) The uncertain model of water quality studied in the dissertation is on the base of deterministic model of water quality. The efficiency of uncertain model of water quality has been increased by the employing ANN. Firstly, Monte Carlo stochastic method is employed to invert the deterministic model to stochastic model; then, the mapping and generalization ability of ANN is also utilized to simulate the solving mechanism of deterministic model of water quality to take the place of huge finite element calculation.(4) The water quality risk of Baiyin section of Yellow River is also evaluated quantitatively by the coupled model and the results suggest that not only the reaching standards but also the risk of exceeding the level of concentration mean of main pollutants all can be estimated by the coupled model of water quality. Since the information furnished by the coupled model of water quality is greatly abundant, it is so useful to control pollution of water quality, forecasting and alarm of accident, protecting water environment and rational utilizing of water resources.(5) The uncertainly coupled model of natural river water quality is compared in detail with present model of water quality by using same example and solving ways.The results of comparing with deterministic model of water quality showed: (1)theaccuracy of uncertainly coupled model of water quality (η|-= 9.6%) is higher than thatof deterministic model of water quality (η|-= 14.8%); (2)the uncertain model of waterquality can not only reflect the deterministically variable trend of concentration as deterministic model but also evaluate water quality risk randomly, so the uncertain model is more versatile; (3)the parameters are more impersonal and rational in uncertain model of water quality, because the main factors influenced water quality variation have been sampled randomly by using Monte Carlo method.The results of comparing with presently uncertain model of water quality showed: (1)the presently uncertain model can only attain the analytic solution of the model equations of regular river water quality, however, the uncertainly coupled model of natural river water quality overcomes the limitation by employing ANSYS; (2)theaccuracy of uncertainly coupled model of water quality (η|- = 9.6%) is higher than that of presently uncertain model of water quality(η|- = 10.6%); (3)the simulation time of theuncertainly coupled model of water quality studied in the paper is only 26.9 percent that of presently uncertain model of water quality by the mapping and generalization ability of ANN, which savings the calculation resources and the efficiency is been increased.2 There are still some problems need an in-depth study:(1) it is necessary for setting up a more perfect and accurate model of water quality to qualitatively and quantitatively research numerously uncertain factors influenced the water quality variation aided by laboratory experiments and on-the-spot monitoring and increase the simulated water quality indicators.(2) in order to test the model comprehensively, the more information of observational data are indispensable.(3) it is necessary for realizing the visual management to integrate the model with GIS to enhance the interaction ability of the uncertain water quality model.