Enhancing Time Series Forecasting with an Optimized Binary Gravitational Search Algorithm for Echo State Networks

The echo state network (ESN) is a cutting-edge reservoir computing technique designed to handle time-dependent data, making it highly effective for addressing time series prediction tasks. ESN inherits the more precise design of standard neural networks and relatively simple learning process has strong capacity solving nonlinear problems. It can disseminate low-dimensional information cues high-dimensional areas enabling extracting data. However, this study proven that not all output dimensions directly impact model generalization. This desires enhance model’s generalization abilities by decreasing redundant feature. A remarkable hybrid proposed optimizes association through feature selection. called binary improved gravitational search algorithm (BIGSA) (BIGSA-ESN). BIGSA’s selection approach complements connection architecture. In study, evaluation was performed using root mean square error (RMSE). experimental findings on Lorenz Mackey–Glass benchmark time-series datasets demonstrate outperforms conventional evolutionary methods. Moreover, empirical predicting significant water quality parameter from wastewater treatment (WWTP) dataset ensemble BIGSA models performs very well in real-world scenarios.