有限信息环境下低压配电系统潮流回溯与动态调度方法研究

The low-voltage power distribution system is the terminal power system that directly supplies power to the users. Due to the lack of SCADA system and GIS system, the low-voltage power distribution system usually operates in a limited information environment, which is characterized by uncertain topology, large randomness of load and renewable energy and incomplete operation state information. This project introduces the Markov random field model to generate the low-voltage distribution system topology based on the user data collected by the smart meter. It also introduces the norm regularization method to improve the calculation efficiency. In order to adapt to the limited information environment of low-voltage power distribution system, this paper proposes a power flow backtracking method to analyze the power flow upper and lower bound, based on which the feasible region of flexible load is optimized and a two-layer scheduling method is constructed. In this way, the uncertainties of load, renewable energies and topologies are considered. On this basis, this project proposes a temporal difference learning based approximate dynamic scheduling method, which dynamically updates the operating state to realize the dynamic scheduling of the low-voltage distribution system. This project will provide theoretical support for promoting demand response, improving electric vehicle and renewable energy consumption capacity, and promoting the development of distribution side power market.

低压配电系统是直接给用户供电的末端电力系统，由于缺乏SCADA系统、GIS系统，低压配电系统通常运行在有限信息环境下，具体表现为拓扑结构不确定、负荷和可再生能源随机性大、运行状态难以获取。本课题通过引入了马尔科夫随机场模型，从智能电表采集的用户数据生成低压配电系统拓扑结构，并引入范数正则化方法提高计算效率。为适应低压配电系统有限信息环境，本课题提出一种基于潮流回溯方法的潮流功率上下界解析方法，并以此优化柔性负荷的可行域，构建一种双层调度方法，实现对负荷随机性和拓扑随机性的管理。在此基础上，本项目提出基于时序差分学习的近似动态调度方法，通过动态更新运行状态，实现低压配电系统的动态调度。本课题将为推动负荷侧响应，提升电动汽车和可再生能源消纳能力和促进配电侧电力市场的发展提供理论支撑。

本课题针对目前低压配电系统所面临的网络量测缺失与终端量测繁荣、用户侧负荷多变与智能化运维缺失等问题，围绕低压配电系统拓扑识别、潮流解析、动态调度方法，提出了基于马尔可夫随机场的低压配电系统拓扑识别方法，实现了有限信息的环境下低压拓扑识别，在此基础上提出了基于低压配电系统潮流回溯法的柔性负荷可行域计算模型，实现了柔性负荷的运行安全裕度计算，最后基于时序差分学习的近似动态调度方法，通过动态更新运行状态，实现了低压配电系统的动态调度。本课题针对用户负荷多变与电网平台不足的特点，利用有限的量测数据实现配电网的智能感知与优化运行，为解决和治理低压配电系统网络损耗、消纳可再生能源、提升用户用电可靠性等一系列问题提供理论支撑和技术支持。