Adaptive particle swarm optimized fuzzy algorithm to predict water table elevation

Abstract

This study helps to select the length for fuzzy sets in fuzzy time series prediction. In order to examine the effect of intervals and evaluate the efficiency of the proposed algorithm, numerical data of water recharge and discharge are considered to predict water table elevation fluctuation (WTEF). Particle swarm optimization (PSO) is an influential tool to handle optimization of multi-model problems, whereas fuzzy logic can handle uncertainty. In this paper, adaptive inertia weights are adopted rather than static inertia weights for PSO, which further improves efficiency of PSO. This modified PSO is termed as adaptive particle swarm optimization (APSO). APSO optimizes the intervals and these intervals are further used to generate fuzzy sets for prediction. The results indicate that the APSO performs better than PSO and genetic algorithm approaches for the same problem.