Unscented Transform based Unbiased Minimum Variance Filter for Generator State Estimation with Unknown Inputs

Dynamic state estimation of synchronous generators is associated with various state space model based monitoring and control strategies. Estimation algorithms based on Kalman Filter are mainly used for estimation of the states locally using the generator input and output measurements. Whereas, wide area power system monitoring and control strategies require remote estimation of states and relies on Phasor Measurement Unit's (PMU) measurement signals which are limited to the generator output measurements like voltage, current, and frequency. In case of basic Kalman Filter based techniques, apart from output measurements, input measurements are also required for the estimation of states. This paper proposes an improved extended Unbiased Minimum Variance Filter algorithm that utilises Unscented Transform for remote state estimation with limited PMU measurements. The estimation is performed in the absence of input measurements. The performance of the filter under variation in input, measurement errors and the effect of initial estimate are analysed.