Water Quality Index Forecast Using Artificial Neural Network Techniques Optimized With Different Metaheuristic Algorithms

Abstract

An accurate water quality index (WQI) forecast is essential for freshwater resources management due to providing early warnings to prevent environmental disasters. This research provides a novel procedure to simulate monthly WQI considering water quality parameters and rainfall. The methodology includes data pre-processing and an artificial neural network (ANN) model integrated with the constraint coefficient-based particle swarm optimization and chaotic gravitational search algorithm (CPSOCGSA). The CPSOCGSA technique was compared with the marine predator's optimization algorithm (MPA) and particle swarm optimization (PSO) to increase the model's reliability. The Yesilirmak River data from 1995 to 2014 was considered to build and inspect the suggested strategy. The outcomes show the pre-processing data methods enhance the quality of the original dataset and identify the optimal predictors' scenario. The CPSOCGSA-ANN algorithm delivers the best performance compared with MPA-ANN and PSO-ANN considering multiple statistical indicators. Overall, the methodology shows good performance with R-2 = 0.965, MAE = 0.01627, and RMSE = 0.0187.