Model-based distributed prediction for contaminant advection diffusion process in liquid flow

In this paper, a model-based contaminant distributed prediction approach based on the combined static nonlinear dynamics and linear dynamics structure is developed to predict the contaminant concentration in environmental fluid diffusion process. Finite element analysis could achieve a near perfect result in computing the contaminant concentration field in predicted water field. However, the finite element method for contaminant concentration diffusion prediction in liquids fluid areas often suffer from inefficiencies due to expensive time-consuming computation, which can't reach the requirement of the real-time prediction. To address the problem, this paper proposed the local model-based contaminant prediction strategy based on linear programming for spatially distributed environmental fluid fields. Then, the three-dimensional partial differential equation of the contaminant concentration is reduced to a linear vector difference equation for local stagnation-point region. The local prediction model is developed to estimate the dynamics of contaminant concentration in local stagnation-point region based on local observed inputs and outputs data. Based on local predicted model, the online correction algorithm is designed to obtain local contaminant concentration prediction outputs through minimizing the local optimization objective function, where the contaminant concentration at a previous time is used to correct the output of local models. Then, the global contaminant concentration is obtained by a linear programming method, where the predictive algorithm performance analysis for the contaminant spatial temporal outputs over the prediction horizon are considered as the optimal objective. Finally, comparison between finite element analysis model and this proposed model validates the accuracy of prediction mode in assessing the effects of contaminant emission on environmental fluid.