Performance guaranteed state estimation for renewable penetration with improved meters

The Journal of Engineering

Ram Rajagopal¹  Jiafan Yu²  Yang Weng³ 

[1] Department of Civil and Environmental Engineering , Stanford University , Stanford, CA , USA;Department of Electrical Engineering , Stanford University , Stanford, CA , USA;School of Electrical, Computer, and Energy Engineering, Arizona State University , Tempe, AZ , USA

关键词: improved meters;    global optimum;    reliable grid operations;    local optimums;    stable grid operations;    extended state space;    short time scale;    measurement noises;    distribution networks;    local search method;    highly accurate smart sensors;    globally optimal solution;    power grids;    measurement noise level;    semidefinite programming-based approaches;    nonconvex;    strong power;    SE problem;    transmission networks;    renewable penetration;    no-measurement-noise assumption;    performance metrics;    state fluctuations;    power system states estimation;    estimation error;   

DOI  :  10.1049/joe.2017.0396

学科分类：工程和技术（综合）

来源: IET

【 摘 要 】

State estimation (SE) aims at monitoring the transmission and distribution networks to achieve stable and reliable grid operations. While the SE problem is non-convex, a local search method is used to achieve global optimum based on the belief that power system states are with limited variation in a short time scale. However, the recent and rapid deployment of renewables leads to strong power and state fluctuations in power grids, making local search method prone to local optimums with large estimation errors. Here, the authors propose to analyse SE with small measurement noises because highly accurate smart sensors are deployed in the past few years. By exploring a special structure of the performance metrics, the authors reformulate the SE problem in an extended state space. The new formulation is convex under the no-measurement-noise assumption. In order to use such a method when there are noises, the authors prove that a perturbation of globally optimal solution is asymptotically bounded by the measurement noise level. This prevents local optimums, which can create a large estimation error. Simulation results demonstrate that the method has a better performance compared to both weighted least square and the recent semidefinite programming-based approaches, especially when the noise is small.

【 授权许可】

CC BY