As the peak-valley difference and load volatility in building energy systems is growing, the traditional cold storage systems that are mainly based on rule-based control or single-layer optimization tend to suffer the lack of foresight, poor constraint adaptability, and slow real time response, and balancing the operational economy and control stability is then difficult. Thus, the paper develops an operation decision-making and adaptive control algorithm in the cold storage systems, combining load forecasting, enhanced intelligent optimization, and model predictive control (MPC) in the complex operating conditions. Feedforward information is made available to optimization in order to provide a shift in response to a forecast-driven one by generating short-term cooling demand sequence by use of rolling forecasts. Addressing the strong constraints, nonlinearity, and multi-peak characteristics of cooling storage scheduling, an improved intelligent optimization solver is designed, incorporating adaptive weights, feasible region repair, and heuristic initialization. The optimization results are coupled with MPC rolling control to construct a hierarchical closed loop of "day-ahead—intraday—real-time". Experimental results show that, compared with traditional rule-based strategies, the peak-segment electricity purchase ratio decreases from 41.3% to 24.7%, the overall system COP increases to 3.56, and the cooling storage utilization rate improves to 86.9%. Simultaneously, the equipment start-up and shutdown frequency decreases to 8.9 times/day, the load tracking MAE decreases to 0.056, the algorithm’s feasible solution rate reaches 99.1%, and the average convergence generation is shortened to 72 generations. The results verify the comprehensive advantages of the proposed method in terms of economy, energy efficiency, and operational stability, providing effective technical support for the intelligent operation of cooling storage systems.
Zhu et al. (Thu,) studied this question.