为解决虚拟电厂聚合分布式资源过程中因风光出力随机、负荷波动及参数偏差等多源不确定性导致的调度失效与经济损失问题，构建了嵌入多源不确定性建模与参数在线校正机制的多时间尺度自适应调度框架。该方法基于两阶段鲁棒优化与改进量子遗传算法，日前阶段通过鲁棒优化生成预调度方案，日内阶段引入状态反馈机制，利用改进量子遗传算法滚动校正关键参数，构建闭环调度结构，并通过仿真试验验证效果。结果表明：在风光出力与电热负荷预测存在显著偏差时，所提方法较传统确定性调度实际运营收益提升约3.2%；参数在线校正策略使系统平衡成本在多数时段显著降低，降幅接近90%。该方法有效协同了调度方案的鲁棒性、经济性与自适应能力，为虚拟电厂在高不确定性环境下的安全经济运行提供了技术路径。

To address the issues of dispatch failure and economic losses caused by multi-source uncertainties—including the randomness of wind and solar power generation, load fluctuations, and parameter deviations—during the aggregation of distributed energy resources in virtual power plants (VPP), this paper proposes a multi-time scale adaptive dispatching framework embedded with multi-source uncertainty modeling and an online parameter correction mechanism. Based on two-stage robust optimization and an improved quantum genetic algorithm (QGA), a pre-dispatch scheme is generated via robust optimization during the day-ahead stage. During the intraday stage, a state feedback mechanism is introduced to rolling-correct key parameters using the improved QGA, thereby establishing a closed-loop dispatching structure. Simulation results demonstrate that under significant prediction deviations in wind/solar generation and electric/thermal loads, the actual operational revenue of the proposed method increases by approximately 3.2% compared to traditional deterministic dispatching. Furthermore, the online parameter correction strategy significantly reduces the system balancing cost in most periods, with a reduction margin approaching 90%. The proposed method effectively coordinates the robustness, economics, and adaptability of the dispatching scheme, providing a technical pathway for the secure and economic operation of VPP in highly uncertain environments.