ESN-based distributed inference methods for production line energy forecasting

The increasing focus on environmental sustainability in industry underscores the importance of accurately predicting energy consumption to achieve efficiency goals. In this context, distributed inference proves to be a promising approach, utilizing the extensive data produced by distributed sensors in industrial environments. This study aims to create a novel methodology for precise energy consumption forecasting by integrating centralized training with distributed inference. The proposed solution has been tested on a real pilot case, a production line provided by a manufacturing company, and the results demonstrate the effectiveness of distributed inference frameworks in promoting industrial sustainability.