计算机科学
调度(生产过程)
国家(计算机科学)
智能电表
实时计算
米
估计
智能电网
测量不确定度
可靠性工程
概率分布
观测误差
分布(数学)
电力系统
工程类
负荷管理
估计理论
作业车间调度
状态信息
可靠性(半导体)
计量单位
控制理论(社会学)
计量系统
数学优化
作者
Wentao Zhang,Yuzhang Lin
标识
DOI:10.1109/tsg.2026.3753578
摘要
The increasing penetration of distributed energy resources has intensified uncertainty in distribution systems, making distribution system state estimation essential for situational awareness and advanced control. Although smart meters improve state-estimation observability, bandwidth limitations often preclude continuous high-rate streaming, and utilities therefore rely on fixed reporting intervals. Such fixed-rate reporting induces asynchronous sampling and sample-and-hold effects, leading to time-skew error and degraded estimation accuracy. This paper proposes a time-skew-error-based sensor-scheduling framework that optimizes the reporting interval of each meter under an explicit communication budget equal to that of fixed-interval baselines. The method models reporting-induced time-skew-error statistics as interval-dependent uncertainty and selects reporting intervals that minimize the resulting error without increasing the total number of uploads. To enhance robustness under nonstationary load and distributed-energy-resource dynamics, the framework integrates distribution-shift detection for periodic re-optimization and event-triggered transmissions to capture photovoltaic-generation and electric-vehicle transients. Case studies on a real-world 127-bus feeder show that, under the same communication budget, the proposed approach consistently reduces the root-mean-square errors of both nodal injection measurements and estimated voltage magnitudes relative to standard fixed-interval reporting across multiple baseline intervals. Under the representative 15-min fixed-interval baseline, the active/reactive-power measurement root-mean-square errors are reduced by 65.18%/66.47% and the voltage-magnitude root-mean-square error by 54.67% in the solar/electric-vehicle-rich scenario; corresponding reductions of 29.37%/33.36% and 27.89% are retained when independently measured 1-min household data are used.
科研通智能强力驱动
Strongly Powered by AbleSci AI