RIA: High-Breakdown Point Estimation in Electric Power Systems

点击翻译按钮获取中文摘要

The security and the economic operation of an electric power system can benefit greatly from robust estimation theory. Its major concern is the design of estimators which are robust against departures from the assumptions, due to bad measurements for instance. The first objective of the proposed research is to extend this theory to electric power systems, which are represented by models involving sparse matrices. In particular, the important concept of breakdown point, which is defined as the maximum fraction of contamination that an estimator can handle, has to be generalized in order to account for the local properties which characterize sparse regression models. Based on this concept, algorithms which apply high-breakdown point estimators to power systems will be developed. Here, the resampling technique and the simulated annealing method will be considered, the objective being the development of algorithms which can be implemented in a real-time environment of a control center. The third objective of the proposed research is the investigation of meter placement methodologies which are optimal for bad data identification. This is indeed an important issue since in power systems the redundancy in the measurements is very low. Preliminary results showed that a poor measurement configuration for bad data analysis can be dramatically improved through appropriate measurement re-allocation.