Robust Identification and Correction of Multiple Parameter Errors in Power Networks Using a Relaxation-Regularization Method

Accurate line and transformer parameters are essential for all application functions in the energy management system. Assessment and decision results will be distorted if parameter errors exist. This paper proposes a novel robust method for simultaneous identification and correction of multiple parameter errors in power networks using a relaxation-regularization technique. The parameter error identification problem is first formulated as a regularized multi-snapshot weighted least squares problem with adjustable constraints on parameter ranges. Erroneous parameters are identified by solving the regularized problem multiple times with increasing regularization parameters. Analytical justification is provided that, by penalizing the parameter ranges through the regularization term, the ranges of correct parameters will rapidly shrink while the shrinkage of erroneous parameters' ranges is much slower. The model is then converted into a robust Huber M-estimation formulation with inequality constraints, which rejects bad measurements automatically. The effectiveness of the proposed method is verified using a large number of erroneous parameter combinations under different measurement noise levels, measurement redundancies and scenarios containing bad measurements with promising results.