Industrial Automatic Control Systems and Controllers Annotation << Back Improving the Energy Efficiency of Gold Enrichment Processes Using Echo-State Networks Litvinov S.N. The purpose of this study is to develop an approach for improving the energy efficiency of gold beneficiation processes using Echo State Networks (ESN), a promising class of reservoir computing. The relevance of the work stems from the need to reduce specific energy consumption at gold processing plants, where technological processes exhibit high inertia, nonlinearity, and transport delays of up to 3–5 minutes. Traditional automation systems based on PID controllers fail to provide optimal control under ore quality fluctuations, leading to excess energy consumption of 10–20 percent. Scientific novelty lies in substantiating the feasibility of ESN for energy-efficient control of grinding, classification, and flotation processes in gold ore processing. We propose a novel hybrid automation system architecture where the ESN acts as an adaptive predictive corrector for base PID loops, with a prediction horizon equal to the transport delay. Methodology includes mathematical modeling of the mill-hydrocyclone-flotation process train, ESN training using ridge regression, and comparative analysis of prediction accuracy and energy efficiency. Simulation results demonstrate a 12–15 percent reduction in mill drive power consumption and an 8–10 percent reduction in flotation reagent consumption, achieved through improved prediction of technological indicators. The normalized root mean square error of hydrocyclone over flow size prediction for ESN was 0.12, twice as good as that of a linear ARXmodel. Practical significance lies in creating an algorithmic foundation for hybrid industrial control systems based on programmable logic controllers supporting embedded artificial intelligence technology. Limitations include the need for careful tuning of reservoir hyperparameters, including reservoir size, spectral radius, and input scaling. Keywords: echo state networks, reservoir computing, energy efficiency, gold beneficiation, automated control systems, industrial controllers, edge AI, hybrid control, time series prediction. DOI: 10.25791/asu.6.2026.1668 Pp. 45-54.