Implicit 4DVar Particle Filter State Estimation of Dynamic Power Systems: Preliminary Results

Abstract

Dynamic state estimation for near real-time applications in power systems is becomingly increasingly important with the integration of variable wind and solar power generation that can be employed even at disaster conditions. New advanced state estimation tools that will replace the old generation must be developed in a general mathematical framework to assess complexity tradeoffs and addressing nonlinearity and non-normal behaviour while exploiting legacy software. Such a framework must also satisfy the power industry requirement for cautious evolutionary change rather than a revolutionary approach. Implicit Particle Filtering (IPF) is a sequential Monte Carlo method for data assimilation that uses an implicit step to select particles from the high-probability region of the implicit distribution. This work develops the formulation of IPF as for the estimation of the states of a power system and presents the first IPF application study on a power system state estimation. The approach is analyzed using a simulation of a three-node benchmark power system. For implicit function four dimensional variational data assimilation is used. The proposed algorithm is also non-intrusive for communications since the algorithm developed will have the flexibility to address multilevel heterogeneous wireless networks in the integration of different data packets.