OPTIMAL ESTIMATION FOR KEY PARAMETERS OF THE MARINE QUALITY MODEL USING DATA-DRIVEN NEURAL NETWORK

Marine water quality models are complicated because of their multi-parameter and multi-response characteristics. One major difficulty with water quality models is the accurate estimation of model parameters. In this paper, a new method based on a data-driven model (DDM) is developed to retrieve the value of model parameters. All training data are calculated by numerical water quality models from results of multi-parameter matching design cases so the physical properties are not disturbed. The concept is to find the relationship between model parameters and the pollution concentration values of interior stations. Field data are imported into the relationship for inversing optimal parameters or near-optimal parameters, ultimately an optimal or near-optimal prediction method is applied to validate the long-term stability of inversion results. Case tests were carried out in the Bohai Sea, China. Chemical oxygen demand (COD), dissolved inorganic nitrogen (DIN), chlorophyll (Chl) and their sensitive parameters were considered for validating the present method. The optimal solution determination method is applied for DIN and Chl owing to existence of the same sensitive parameters. Case studies show that the present method can make a more satisfactory estimation for this practical problem.