Optimizing Smart Home Energy Forecasting: Unveiling Superiority among RBFNN, ESN, and LSTM

In response to the accelerating integration of IoT devices for monitoring residential energy consumption, this study focuses on identifying robust predictive models aligned with global sustainability goals. Investigating three advanced models-Radial Basis Function Neural Networks (RBFNN), Echo State Networks (ESN), and Long Short-Term Memory (LSTM)-reveals distinct performance metrics. RBFNN demonstrates solid performance, while after meticulous tuning, ESN exhibits superior predictive capabilities, outperforming traditional techniques and even LSTM. Employing Recursive Feature Elimination (RFE) identifies key predictors, emphasizing ESN's heightened predictive accuracy. This research provides empirical insights, highlighting ESN's potential in enabling sustainable energy management within smart homes.