针对短流程炼钢园区综合能源系统日内运行中高比例新能源与生产负荷时序错配，导致的消纳水平与经济性受限问题，提出了基于模型预测控制(model predictive control, MPC)的炼钢园区综合能源系统多时间尺度日内调度策略。首先，通过分析能耗与工序时序特性的关系，构建考虑工序顺序、工序间等待与任务时间平移约束的钢铁生产调节模型，揭示短流程炼钢多工序耦合的用能特性；其次构建双层多时间尺度MPC日内调度策略，以日内运行总成本最小为目标，上层以1h分辨率统筹机组启停、储能与生产任务安排，下层以15min分辨率滚动修正机组出力与储能充放电以补偿风光与负荷预测误差，实现实时功率平衡。最后，以高比例风光接入的工业园区算例验证所提方法，并与固定日前调度及单层MPC进行对比。结果表明：所提双层多时间尺度MPC能够在新能源不确定性与负荷波动条件下实现滚动校正与实时平衡，显著提升运行经济性与新能源消纳能力，有效降低弃风弃光水平。

To address the temporal mismatch between high-penetration renewable energy and production loads in short-process steelmaking park integrated energy systems, which constrains renewable integration and economic performance during intraday operations, this paper proposes a multi-timescale intraday dispatch strategy based on model predictive control (MPC). First, by analyzing the relationship between energy consumption and process timing characteristics, a steel production regulation model is developed incorporating process sequencing, inter-process waiting time, and task shifting constraints, revealing the coupled energy consumption characteristics of multi-process short-process steelmaking. Second, a bi-level multi-timescale MPC intraday dispatch strategy is constructed to minimize total intraday operating costs: the upper layer coordinates unit commitment, energy storage, and production scheduling at 1-hour resolution, while the lower layer performs rolling correction of unit output and storage charging and discharging at 15-minute resolution to compensate for wind, solar and load forecast errors, achieving real-time power balance. Finally, the proposed method is validated through a case study of an industrial park with high renewable penetration and compared with fixed day-ahead scheduling and single-layer MPC. Results demonstrate that the proposed bi-level multi-timescale MPC achieves rolling correction and real-time balance under renewable uncertainty and load fluctuations, significantly improving operational economy and renewable integration while effectively reducing wind and solar curtailment.