小时级月度发购电计划的优化模型及其混合智能求解策略

Optimization Model of Hour-Level Monthly Generation and Purchase Scheduling and Its Hybrid Intelligent Solution Strategy

结合网内外购电模式的差异,建立了计及网外购电市场的小时级月度发购电计划模型。该模型是一个多时段大规模混合整数规划模型。负荷的时段融合策略可有效减小模型中小时级时段数的规模,但会带来负荷融合误差,从而影响模型的目标甚至安全约束,为此提出了基于负荷融合误差的校正优化策略。针对模型的大规模多时段混合整数规划特点,提出了结合免疫遗传算法与拉格朗日分解协调策略的混合智能算法。一方面,利用免疫遗传算法为拉格朗日松弛法提供满足启停约束和旋转备用约束的机组组合发购电方案。另一方面,利用拉格朗日分解协调算法高效求解运行机组的最优发电方案,快速提升免疫遗传算法优秀个体的质量。两者交替迭代,互相促进,从而实现上述多时段混合整数规划模型的高效求解。最后通过IEEE 57节点系统的仿真分析验证了所提模型和算法的有效性。

Based on different electricity purchase modes of internal and external grids, an optimization model of hour-level monthly generation and purchase scheduling is proposed, considering external electricity purchase market of a multi-period, large-scale and mixed integer programming model. Although load partition technology can reduce scale of the hour-level periods effectively, it brings load fusion error impacting objective and even security constraints. Thus correction optimization strategy based on load fusion error is proposed. Because the model is characterized with multi-period,large-scale and mixed integer programming, a hybrid intelligent algorithm combining immune genetic algorithm（IGA） and Lagrange relaxation decomposition-coordination strategy（LR） is proposed. IGA provides LR with unit commitment and generation-purchasing scheme, satisfying unit start-stop and spinning reserve constraints. LR efficiently solves optimal generation scheme of operating units, improving quality of excellent individuals among IGA quickly. Through alternate iteration and mutual promotion of IGA and LR, the proposed multi-period mixed integer programming model can be solved efficiently. Finally, IEEE 57-bus system is used as an example to test effectiveness of the proposed model and algorithm.