Multi-media energy forecasting and optimization serve as critical prerequisites for achieving energy balance and efficient operation in steel energy systems. However, influenced by complex processes and human factors, energy production and consumption exhibit inherent uncertainty, limiting forecasting accuracy. Concurrently, energy optimization methods directly determine the rationality of energy scheduling strategies. To enhance the accuracy of energy production and consumption forecasts and improve the effectiveness of subsequent optimization strategies, this paper proposes a multi-media optimization method for steel energy systems based on operational condition variations. First, a multi-energy media forecasting approach adapted to changing conditions is introduced. To fully leverage production planning information and account for the impact of variable conditions on energy balance, the dynamic time warping barycenter averaging method is employed to identify energy fluctuations. This tracks and corrects operational conditions of energy equipment, providing energy fluctuation forecasts under combined operational conditions. Building upon this foundation, the method comprehensively considers process conditions affecting energy scheduling. Accurate modeling of energy units is achieved using radial basis function neural networks, establishing an energy optimization model. This model is integrated with a rule-based energy allocation approach to enhance the real-time efficiency and robustness of the optimization methodology. To validate the effectiveness of the proposed method, experiments were conducted using actual energy data from a domestic steel enterprise. The results confirm the proposed method delivers effective outcomes in the optimized scheduling of multi-energy media within steel production.
Sheng et al. (Fri,) studied this question.