A method for water pollution source localization with sensor monitoring networks

Abstract

Research on pollution localization using sensor monitoring networks has important significance for environmental protection. There are some challenges in the detection and localization of water pollution sources due to the particularity and complexity of the water environment. The parameters of the traditional concentration diffusion model cannot be obtained in real time, making it difficult to apply. Complex and uncertain interference, such as time delays, flow velocity, and the fact that propagation relies heavily on propagation distance, makes the positions of pollution sources not constant, and the sensor data cannot reach the information processing center at the same time. Therefore, a nonlinear concentration diffusion model formulated by combining Extreme Learning Machine (ELM) with Partial Least Squares (PLS) has been established to describe the relationship between the concentration diffusion value and the position of the sensor in order to solve the problem that the traditional model parameters cannot be acquired in real time. Furthermore, the Sequential Unscented Kalman Filter (SUKF) combined with the ELM-PLS model is utilized to realize precise localization. These proposed methods could achieve accurate diffusion modeling and localization by addressing any nonlinear diffusion data. This would overcome the existing shortcomings of the standard PLS modeling method and LS localization method. Finally, the effectiveness of these proposed methods are verified by numerical simulation examples.