A study on the inverse identification method of line parameters in distribution networks by fusing SCADA and μPMU data

To enhance the accuracy and robustness of line parameter identification in distribution networks, this study proposes a unified reverse identification framework that fuses SCADA and μPMU data. A non-convex optimization model is formulated with the objective of minimizing multi-source measurement bias using the L1 norm, and efficiently solved by the differential evolution (DE) algorithm. In simulations of 110 kV and 35 kV networks with Gaussian noise, the proposed method achieved average relative errors of 0.58% for resistance and 0.37% for reactance, outperforming GA (0.62% and 0.45%) and PSO (0.72% and 0.51%) while reducing computation time by over 40%. Importantly, the study revealed the “low identifiability” of susceptance, where errors exceeded 30% even under moderate noise, and verified that a higher R/X ratio improved robustness in resistance and reactance identification. These findings highlight both the effectiveness of the proposed framework and the structural challenges in parameter identification, providing concrete directions for future algorithmic improvements.