Novel resource-efficient algorithms for state estimation in the future grid

Abstract

The purpose of this study is to examine the challenges and opportunities presented by the evolution of the legacy grid into a smart grid within a framework relevant to signal processing research. The focus here is on the application of statistical signal processing techniques to design new algorithms for state estimation which is a key function in the supervisory control and planning of power grids. To begin with, two resource-efficient forecasting-aided state estimation algorithms are developed which can combine measurements from both the traditional measurement devices as well as the newer measuring devices known as phasor measurement units and operate on them optimally in order to arrive at an estimate of the system state. The ability of the algorithms to track the evolution of the state vector in time is verified using a computer simulation and their statistical performance with respect to the root mean-squared error is studied. Since concentrating the state estimation function at a single point to monitor a large interconnection leads to huge communication and computational overhead, a more feasible approach is to distribute the state estimation function throughout the interconnection. An on-demand, distributed estimation scheme which features event-triggered communication is developed herein to reduce the communication and computational overhead associated with distributed estimation in large systems. This technique is derived from a cutting-edge signal processing paradigm known as set-membership adaptive filtering. The performance of the new algorithm is studied using computer simulations and comparisons are drawn to existing adaptive filtering methods like the recursive least-squares method.