A 3D Robotic Inspection Framework for Power Quality Monitoring in Subterranean High-Voltage Transmission Systems

The operation of underground power transmission systems is based on continuous monitoring to ensure reliable power transmission, maintain power quality, and avoid unexpected failures. But most of the existing robotic inspection systems are mainly based on navigation and visual inspection, which have limited ability in integrating power quality monitoring and intelligent fault analysis. To overcome these drawbacks, this work is proposing an intelligent robotic inspection framework that combines autonomous robotic monitoring, multimodal sensing, optimized deep learning and adaptive hyper-parameter optimization for power quality disturbance detection. The proposed framework uses a 101-layer Squeeze-and-Excitation Residual Network (SE-ResNet-101x) to classify pre-processed one-dimensional voltage signals into Normal and Fault classes and hyperparameters are optimized through the Self-Adaptive Sea Lion Optimization (SA-SLnO) algorithm. A set of data is generated based on simulation, to mimic actual underground inspection scenario, including the electrical, environment and robotic operational factors. The LiDAR data is used in isolation to provide 3-dimensional localization for the robot while the thermal and camera data provide additional inspection data.Experimental results show that fault detection performance is very good, with 99.10% accuracy, 99.97% precision, 98.94% recall and 99.45% F1-score. The proposed system has been able to successfully detect voltage sags, voltage swells, Harmonic distortions, Transient disturbances and Flicker events. Moreover, high climbing efficiency of more than 89 % is achieved on slopes of 30°. Even in wireless mode, high communication reliability of 99.8 % is achieved. The proposed framework is an effective way to monitor underground power automatically, which can achieve accurate early detection of fault and reliable underground power monitoring inspection by robots in complex underground power transmission scenarios.