Optimization of Deep Reservoir Computing with Binary Genetic Algorithm for Multi-Time Horizon Forecasting of Power Consumption

The roll of consumption energy forecasting is very important to make planning time-horizon strategy, and mitigate a great management. As result, improving the sustainability energy, creating clean environment. Aiming develop in different time horizons, this work gives results new hybrid method, which combine deep echo state network (DeepESN), with Binary genetic algorithm (BGA). DeepESN an extension Echo (ESN), integrates strong nonlinear series processing capability (of ESN) advanced learning characteristic models. BGA another version optimization methods that can be applied find best values architecture hyperparameters models, basesd on binary decoding his chromosoms. In work, we compared accuracy performance proposed model DeepESN-BGA other methods. It found have fast addition, it based error metrics, without BGA, horizon forecasting. Proposed has been also DeepESN-DE, DeepESN-GA, DeepESN-PSO aiming evaluate term optimization. statistically good result