The article considers the state and developments of sustainable manufacturing systems, focused on improving their energy efficiency. A brief overview of possible production management mechanisms based on machine learning is provided. A three-level system for managing energy-efficient manufacturing in an enterprise is being researched, with the owner at the top level, the manager at the middle level, and the manufacturer at the bottom. It is required to ensure the maximum volume of manufacturing in an enterprise with: (i) given energy resources, (ii) uncertainty due to random external factors, and (iii) undesirable activity of the enterprise's employees - the manager and the manufacturer, associated with the presence of their own goals that do not coincide with the goal of the owner. Organizational mechanism using a supervised learning algorithm formed to control the manager by the owner. Interests and decisions of the manager at this organizational mechanism are analyzed. In addition, similar organizational mechanism with supervised learning algorithm is developed to control the manufacturer by the manager. Interests and decisions of the manufacturer at this organizational mechanism are identified. It is proven that the symbiosis of these two control mechanisms maximizes total manufacturing with available energy resources and increases the energy efficiency of the enterprise.