智能配电网微型同步相量测量单元优化配置及鲁棒状态估计研究

The basis for the smart distribution network to achieve efficient control and optimization lies in the advanced measurement devices and the accurate state estimation of power grid. In recent years, our country has been focusing on the development of micro-synchronous phasor measurement unit (μPMU) which could be deployed on a large scale. However, the optimal placement methods and the state estimation approaches regarding to μPMU are not perfect. This project focuses on three aspects: 1) Develop an optimization method that could obtain all optimized placement schemes, use the robust statistics theory and influence function to deduce the state estimation variance under the Gaussian or non-Gaussian noise assumptions, and reveal the constitutive relationship between the estimation effect and the placement as well as the precision parameters of μPMU. 2) For the measurement uncertainty and data loss occurring in practice, the t-distribution noise and the Variational Auto-Encoder are introduced into the distribution network to build up the uncertain measurement model and restore the missing data respectively. The robust state estimation method is established and analyzed theoretically to reveal the intrinsic relationship between the mean square error of estimation results and the setting parameters of the proposed algorithm and then to improve the accuracy and performance of state estimation. 3) For large-scale smart distribution network, we explore the distributed robust state estimation method and its convergence theory so as to reduce computation load and communication load and eventually save costs. This research is in line with the development of smart distribution network, and the research results will provide the theoretical basis and technical support for the determination of final optimized placement scheme, the optimization and upgrade of the measurement system and the implementation of the distributed robust state estimation method.

智能配电网实现高效控制和优化的基础在于先进量测装置和对全网状态的准确估计。近年来国家重点研发可大规模部署的微型同步相量测量单元（μPMU），但基于μPMU的优化配置和状态估计方法还不够完善。本项目重点研究三个内容：1)发展可获得全部优化配置方案的优化方法，利用稳健统计理论和影响函数推导高斯/非高斯噪声下的状态估计量方差，揭示估计效果与μPMU配置布局、精度参数的本质关系；2)针对现场出现的量测不确定性和数据丢包，将t分布和变分自编码器引入配电网，建立量测不确定模型和重构丢失数据，构建鲁棒状态估计方法并理论分析，揭示均方误差与算法设置参数的内在联系，提升估计精度和性能；3)针对大规模智能配电网，探索分布式鲁棒状态估计方法及其收敛依据，降低计算量和通讯量，节约成本。本研究符合智能配电网发展趋势，研究成果将为最终优化配置方案确定、量测系统优化升级、分布式鲁棒估计算法实施提供理论依据和技术支持。

针对智能配电网，近年来国家重点研发可大规模部署的微型同步相量测量单元（μPMU），但基于μPMU的优化配置和状态估计方法可进一步完善和优化。首先，项目组采用稳健统计理论和影响函数推导非高斯噪声模型下的状态估计量方差表达式，将状态估计量方差作为μPMU优化配置方案的评价指标之一，揭示智能配电网状态估计效果与μPMU配置方案、精度参数之间的本质数学关系和变化趋势，建立了一套μPMU优化算法。其次，项目组采用可灵活模拟高斯或非高斯噪声的t分布来构建量测不确定模型，利用变分自编码器重构丢失数据，进而根据最大似然估计理论，构建基于t分布的鲁棒状态估计方法，以增强抗干扰能力和提高估计精度。进一步，针对大量分布式能源接入的智能配电网规模与复杂性日益扩大的问题，项目组采用分布式优化、图论、最大似然估计等理论方法，发展一套基于t分布的分布式鲁棒状态估计方法，可有效减少计算量和通讯量。最后，在完成项目计划的基础上，项目组结合学术研究发展动向，对研究课题做了一定程度的延伸，开展了电力系统鲁棒动态状态估计方法的研究。本项目研究符合智能电网发展趋势，从理论层面提炼μPMU优化配置算法和量测系统优化升级的理论依据，从应用层面设计集中式和分布式鲁棒状态估计方法。项目研究成果对μPMU的大规模部署和鲁棒状态估计方法的实施具有积极的理论指导意义和应用价值。