River Health Assessment Based on an Artificial Neural Network

Abstract

River ecological health evaluation is a complex problem, and although there are many existing river evaluation methods, it is still a challenge to synthesize them and better express the nonlinear relationships among multiple influencing factors of river ecosystems. In this paper, we try to evaluate the ecological health of Dagu River using artificial neural network, and train and validate the designed network to establish the corresponding river health evaluation model of Dagu River. The analysis of the results shows that the mean absolute error (MAE) of the test data is 4.98, and as a comparison, the MAE is 8.125 when multiple linear regression is chosen as the baseline algorithm, indicating that the artificial neural network can accurately evaluate the river health condition of Dagu River. By using artificial neural network to evaluate the river health, the method has wide practicability and provides the basis for the credibility of river health.