A Simplified Bayesian Learning Technique for Harmonic State Estimation

Direct harmonic monitoring of a complete power system can be costly and impractical. Harmonic State Estimation (HSE) refers to indirect monitoring techniques, where unknown harmonic variables are estimated based on limited observations. HSE is crucial in developing harmonic monitoring systems and thus enabling high power quality. In this paper, a universal formulation for HSE is derived and a Simplified Bayesian Learning (SBL) technique based on Markov Chain Monte Carlo (MCMC) simulation is proposed to solve the problem for different cases of simultaneously operating harmonic sources. Metropolis Random Walk (MRW) and Importance Sampling (IS) are used for MCMC sampling, where the latter can be used to alleviate the problem of needing to choose the proposal distribution. Different cases in the presence of uncertainty in measurements and network parameters are studied, demonstrating the usefulness of the proposed method.