Unbiased Euclidean Direction Search Algorithm for Partial Discharge Location in Cable Systems

Utilizing the unbiasedness criterion, this article proposes a bias-compensated normalized Euclidean direction search (BC-NEDS) algorithm with noisy inputs, which can effectively mitigate the impact of the bias by estimating the statistical properties of inputs. Moreover, the theoretical analysis of the BC-NEDS algorithm is conducted in a transient regime. Simulations verify the validity of the theoretical analysis and showcase the improved performance of the BC-NEDS algorithm. Partial discharge (PD) location techniques can be utilized to effectively monitor the condition of the electrical apparatus. Considering the performance of the conventional location algorithms based on the time difference of arrival (TDOA) method may notably degrade with noisy inputs. Hitherto, scarce literature concentrates on the PD location problem by leveraging adaptive filtering techniques. Such problem is essentially characterized by a noisy input model. This work aims to propose a new one-step algorithm to simultaneously denoise and achieve improved location accuracy with noisy input. The BC-NEDS algorithm is employed for solving the PD location problem. The BC-NEDS algorithm demonstrates the effectiveness in mitigating the bias arising from noisy inputs in both the direct and reflected PD signals and estimating the time difference to locate the PD signal of cable systems. Simulations and experimental studies exhibit that the BC-NEDS algorithm verifies the effectiveness and achieves enhanced location accuracy for cable systems.