Water quality evaluation model based on hybrid PSO-BP neural network

Abstract

A hybrid neural network algorithm, aims at evaluating water quality, based on particle swarm optimization (PSO) algorithm, which has a keen ability in global search and back propagation (BP) algorithm that has a strong ability in local search. Heuristics has been proposed to optimize the number of neurons in the hidden layer. The comparison with the traditional BP NN shows the advantage of the proposed method with high precision and good correlation. The values of average absolute deviation (AAD), standard deviation error (SDE) and squared correlation coefficient (R²) are 0.0072, 0.0208 and 0.98845, respectively. The results show that the hybrid PSO-BP NN has a good predictal ability of evaluating water quality; it is a practical and efficacious method to evaluate water quality.

DOI : http://dx.doi.org/10.11591/telkomnika.v12i2.3190