Predictive Control of an Intelligent Energy-saving Operation System Based on Deep Learning

Abstract

An intelligent energy-saving operation system is a high-tech product specifically designed to transform the air conditioning systems, motor systems, and lighting systems, to reduce energy consumption. The concentration of equipment distribution within these systems leads to a strong coupling relationship between them. By conducting an overall energy efficiency prediction, the intelligent energy-saving operation system can fully explore its energy-saving potential. The existing research methods for the online control process of intelligent energy-saving operation systems are not accurate enough to predict energy-saving operations when numerous devices are involved. Consequently, this article focuses on studying the predictive control of an intelligent energy-saving operation system using deep learning techniques. The Generalized Regression Neural Network (GRNN) network is selected to describe the energy consumption of the system. The Beetle Antennae search algorithm is then employed to iteratively optimize the smoothing factor of the model, eliminating the need to rely on experiential parameter determination and enhancing the predictive performance of the model. For the predictive control of the intelligent energy-saving operation system, the optimized GRNN network model serves as the prediction model. The primary control objective is to minimize energy consumption while maintaining a unified carrying capacity, thus achieving intelligent energy-saving effects. Experimental results validate the effectiveness of the model. ACM CCS (2012) Classification: Computing methodologies → Machine learning → Machine learning approaches → Neural networks Computing methodologies → Artificial intelligence → Control methods → Computational control theory.