适应自律协同调控需求的主动配电网量测配置与分布式状态估计

受端电网中本地传统电源被特高压输电取代和新能源占比进一步提升使得电力系统可控资源不足，需要配电网主动提供调控资源，而主动配电网参与调控的挑战之一是满足调控需求的可观性缺失。调控资源分散难以集中调控和增量配电网部分有保护隐私需求使得主动配电网需采用分散自律协同调控模式，且不同区域在调控中功能不同，对可观性要求不同。本项目拟突破传统系统可观性定义的局限性，提出适应主动配电网自律协同调控需求的可观性与量测配置理论，揭示量测配置对可观性和分布式状态估计计算性能的影响机理，从模型、算法与优化量测配置三个角度提升状态估计性能。为解决以上关键科学问题，拟研究：(1)考虑智能终端部署成本，研究基于知识蒸馏的伪实时量测生成方法；(2)研究适应自律协同调控需求的可观性表征和分布式量测优化配置方法；(3)研究基于数据机理融合驱动模型的分布式坏数据辨识方法；(4)研究适应换流器非光滑特性的分布式状态估计方法。

In the receiving-end power grid, the replacement of local traditional power sources by UHV transmission and the further increase of the proportion of renewable energy sources make the regulation resources of the power system insufficient, which requires the distribution network to actively provide regulation resources. One of the challenges for the active distribution network to participate in regulation is the lack of observability to meet the regulation demand. The decentralized control resources are difficult to centralized control and the incremental distribution network has privacy protection requirements, so the active distribution network needs to adopt decentralized autonomous collaboration control mode, and different regions have different functions in the scheduling and control, and have different requirements for observability. This project intends to break through the limitations of the traditional definition of observability of the system, propose the observability and measurement placement theory to adapt to the demand of autonomous collaboration scheduling and control of active distribution network and reveal the influence mechanism of measurement placement on the observability of active distribution network and the calculation performance of distributed state estimation. The state estimation performance will be improved from three perspectives including model, algorithm, and optimized measurement placement. In order to solve the above key scientific problems, the following issues are studied: (1) the pseudo-real-time measurement generating method based on knowledge distillation, considering the deployment cost of intelligent terminals; (2) the observability representation and distributed measurement placement optimization method that adapt to autonomous collaboration scheduling and control; (3) the distributed bad data identification method based on the data-mechanism fusion driven model;(4) the distributed state estimation method that adapts to the non-smooth characteristics of the converters.