A Data-driven Technique for Network Line Parameter Estimation Using Gaussian Processes

This paper proposes a unique data-driven physics-informed approach to network line parameter estimation, where the parameters of linear continuous-time domain equation governing line dynamics are learnt by modeling the line end voltage and current signals as Gaussian processes. The proposed method allows parameter estimation along with prediction of measurement signal and associated uncertainty in a single framework. The method is tested for parameter estimation on IEEE 14-bus and 9-bus network under steady-state operating condition.