基于协同一致性迁移Q学习算法的虚拟发电部落AGC功率动态分配

发布时间：2018-03-03 19:18

本文选题：一致性迁移Q学习　切入点：虚拟发电部落　出处：《中国电机工程学报》2017年05期 　论文类型：期刊论文

【摘要】：为适应智能电网分散自治的发展趋势,文中在虚拟发电部落控制框架下,提出了一致性迁移Q学习的AGC功率动态分配方法。通过构建两层的功率分配模式,有效解决了机组规模较大导致的"维数灾难"问题。在每个部落与其相邻部落进行值函数矩阵的交互一致性计算后,部落领导者能自组织地协同各个部落的发电功率,从而达到"分散自治,集中协调"的效果。在引入迁移学习后,算法能有效地利用历史优化信息进行快速的功率动态分配优化,以满足AGC的控制时间尺度要求。广东电网模型仿真表明:与集中式分配算法相比,文中所提算法能有效解决复杂大规模电网AGC功率动态分配的分散式优化问题,在减少AGC机组调节费用的同时,可以提高区域电网的控制性能标准。
[Abstract]:In order to adapt to the development trend of decentralized autonomy in smart grid, a dynamic AGC power allocation method based on uniform migration Q learning is proposed under the framework of virtual generation tribal control. The problem of "dimension disaster" caused by the larger unit size is solved effectively. After each tribe and its neighboring tribes have the interactive consistency calculation of the value function matrix, the tribal leaders can organize themselves to cooperate with the generation power of each tribe. Thus, the effect of "decentralized autonomy, centralized coordination" can be achieved. After introducing transfer learning, the algorithm can effectively use historical optimization information to optimize power dynamic allocation rapidly. In order to meet the control time scale requirements of AGC, the simulation results of Guangdong power network model show that compared with centralized allocation algorithm, the proposed algorithm can effectively solve the decentralized optimization problem of AGC power dynamic distribution in complex large-scale power grid. At the same time, it can improve the control performance standard of the regional power network while reducing the regulation cost of the AGC unit.
【作者单位】：广东省绿色能源技术重点实验室(华南理工大学);
【基金】：国家重点基础研究发展计划(973计划)(2013CB228205) 国家自然科学基金项目(51177051,51477055)~~
【分类号】：TM73;TM76

【相似文献】