An edge computing system for live-line detection based on multi-sensor data fusion

As smart grid initiatives gain momentum, higher demands have emerged for real-time, precise sensing of power equipment operational status.Traditional live-line detection methods predominantly rely on single sensors, suffering from issues such as low detection efficiency and poor real-time performance.To address these challenges, this paper first designs an edge computing system tailored for live-line data detection in smart grids.Through multi-sensor data fusion, edge nodes can form a collaborative network capable of more rapidly identifying the location of live-line faults.Then, a dual-channel convolutional neural network is designed to accommodate multi-sensor fusion data inputs.A time-frequency attention feature fusion module is incorporated to optimise the feature extraction process.Experimental results demonstrate that the proposed system achieves live detection within 6.3 milliseconds, with a detection accuracy rate of 94.61%.This system enables timely and precise identification of potential equipment faults while maintaining high real-time performance.