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.