Intelligent Simulation of Water Environmental Pollutant Flux in River Basins

Abstract

In order to realize the simulation of water environmental pollutant flux of different cross sections in rivers basins with storm water management model (SWMM), an intelligent pollutant flux simulation method based on the LSTM-PSO hybrid algorithm is proposed. This method is focused on solving two problems, the data discretization of online water quality monitoring and the complicated calibration of model parameters. Taking typical urban rivers in the eastern region of China as the research object, the experimental analysis is carried out. After the basic SWMM model construction of the study area, long short-term memory (LSTM) neural networks are used to fit the discrete data of online water quality monitoring, and particle swarm optimization (PSO) is used to optimize the selection of model parameters. The pollutant flux in the river basin is simulated, and the simulated pollutant flux of the key cross section is compared with the measured pollutant flux. The experimental results show that the LSTM-PSO algorithm obtains higher accuracy than the conventional approach in the pollutant flux simulation and its efficiency is verified.