To improve the reliability of the electrical grid, its weak points must be identified, and once these are selected, tools and methods for continuous monitoring of the identified elements must be made available. Among the components most prone to failure are the joints of underground cables. These joints are often made in cramped environments where dusty materials are present, which can inadvertently be introduced into the joint during assembly, creating a defect in the structure that could lead to failure. A possible failure can result in a power outage, causing significant technical and economic damage to the power distributor.

The dielectric breakdown of joints is often caused by impurities inadvertently introduced during installation, which lead to partial discharges within the dielectric material. Gradually, these discharges erode the insulating material between the central conductor and the metallic shield. This erosion process is known as “treeing.” Once the defect penetrates the entire layer of insulating material, the cable is subjected to a destructive discharge that takes the entire cable out of service. Currently, the diagnostic methods available on the market (both sensors and algorithms) do not allow for continuous monitoring of partial discharges in cables, nor their precise localization. Consequently, the progression of the defect over time cannot be tracked.

The purpose of this work is to provide a method and a sensor for defect localization in cable joints. This methodology involves continuous monitoring, enabling the tracking of defect evolution over time. This aspect is crucial for estimating the health status of the joint and defining a strategy for replacing defective joints, thereby preventing cable failures. To this end, the algorithm and measurement system were validated on two real joints, produced in collaboration with Unareti S.p.A.: one was made to standard specifications, while the other, with a known defect, was subjected to mechanical stress. Both joints were powered and continuously monitored for four months each. The results demonstrate that the developed sensor and algorithm allow for the measurement of partial discharges due to joint defects and enable the precise localization of these defects. These results were further confirmed by visual inspection of the joint after testing, where the dielectric material erosion caused by partial discharges was visually observed at the location indicated by the defect localization.