Optimal Design of Sensor Arrays in Electromagnetic Tomography for Non-Destructive Treeing Detection in XLPE Cables

XLPE cables are prone to insulation aging faults such as water trees and electrical trees during long-term operation. Once treeing defects occur, they could lead to discharge or breakdown in XLPE cable, seriously threaten the safe and stable operation of the power grid. Accurate positioning and detection of treeing defects has become an urgent need for cable condition monitoring. Electromagnetic tomography (EMT) has the advantages of non-invasive, non-contact, and permeable through metal media, and is expected to be applied in non-destructive testing of treeing faults in XLPE cables. The precision of EMT system highly relies on the hardware design of sensor arrays. To explore the exact relationship between structural design of sensor arrays and the quality of electromagnetic tomography imaging, this paper optimizes the sensor array by changing the number of coils in the sensor array and modifying the traditional coil structure to a coaxial differential structure. The sensor array with different structures is modelled in COMSOL Multiphysics finite element software, and Tikhonov algorithm is used to solve the inverse problem of internal space reconstructions in XPLE cables by EMT imaging. Treeing defects with different sizes and locations are modelled and the imaging effects are compared. Research results indicate that by increasing the number of sensors, imaging quality can be improved accordingly. The usage of coaxial differential structure can eliminate the influence of excitation magnetic field and improve the accuracy of image reconstruction for treeing defects. Research work in this paper has can be a good reference for the setting of sensor arrays in the next step of EMT system design.