Localisation of Partial Discharge Source in Oil Insulation Using Acoustic Emission Technique non Iterative Method Newtons Method and Genetic Algorithm

Abstract

The power transformers are a vital component of power systems. The condition assessment of transformers is of utmost importance to ensure the reliable operation of the power system. The partial discharges (PD) originating from defects in operating transformers should be detected as early as possible. In large power apparatus like transformers, locating the source of PD is as important as identifying it. The PD source localisation helps in risk assessment and in planning of maintenance activities.The acoustic emission (AE) technique is one of the on-line non-destructive testing (NDT) techniques for PD source localisation in power transformers. The PD source is located by solving a system of non-linear sphere equations obtained by modeling the acoustic emission partial discharge (AEPD) location system mathematically. The algorithms that have been developed to solve the mathematical model of AEPD location system need to be improved due to the various limitations. Hence, the current research proposal aims to address this existing research gap by suggesting new algorithms/modifications in the existing algorithms. Further, the factors which affect the accuracy of PD source localisation will be studied and analysed.According to IEEE standard C.57.127-2007, there are two AEPD location systems: (i) all acoustic system; and (ii) combined acoustic-electrical system. When AE technique is used for PD source localisation in power transformers, the error in PD localisation can occur mainly due to two reasons: (i) the inefficacy of the algorithm used for solving the mathematical model; and (ii) the error in measurement of acoustic signal arrival time from the PD source to various sensors.For an all-acoustic system, a hybrid method combining the advantages of both the iterative and random search algorithms is developed to solve the mathematical model of AEPD location system. The existing non-iterative algorithm is modified/extended so that it works for caseswith zero time-differences.