Weak faults in high-voltage cable systems present a persistent challenge for grid operators: they release insufficient energy to trigger standard protection relays, yet they can silently evolve into permanent failures that disrupt power delivery. Common scenarios include jacket degradation from environmental stress and partial electrical breakdown between the cable core and its conductive sheath layer.
A new diagnostic approach addresses this blind spot by monitoring the three-phase sheath currents that naturally circulate through cross-bonding boxes—the hardware joints that connect cable sheaths at intervals along underground transmission routes. Rather than relying on fault signatures from historical incidents, the method employs a statistical distance metric called the Mahalanobis distance to recognize abnormal patterns in current topology.
The technique operates without complex signal decomposition or large training datasets. Instead, it establishes a baseline healthy reference from normal operation data, then flags deviations in the joint structure induced by weak faults. The researchers confirmed the concept on a detailed electromagnetic model of a 110 kV system, successfully identifying all 52 synthetic fault cases within 6 milliseconds while remaining immune to false alarms during healthy operation tests.
Key strengths include robustness under noisy measurements, applicability to asymmetric cable layouts, and modest computational cost per sample—essential for continuous monitoring in field deployments. Boundary-case faults under suboptimal joint conditions still exhibited a nine-fold safety margin, and Monte Carlo simulations across multiple noise levels produced zero spurious alarms.
This advance addresses a critical gap in cable condition monitoring. By detecting incipient faults weeks or months before catastrophic breakdown, utilities can schedule planned maintenance, reduce unplanned outages, and extend cable asset life. The method's requirement for only a single measurement point per cross-bonding location also simplifies sensor infrastructure compared to distributed multi-point monitoring schemes.
