Skip to content
Log in
Sponsored AI Networking Cookbook: Practical recipes for AI-assisted network automation and development AI Networking Cookbook: Practical recipes for AI-assisted network automation and development

Author

Guo-Bang Ban

h-index 4 View on Semantic Scholar

We have 3 of 27 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Review Open access 2026

Helmet-Aware Identity-Consistent Tracking for Safe and Intelligent Power-System Operation and Maintenance

: Safe and reliable operation of critical energy infrastructure requires continuous situational awareness of personnel during substation inspection, power-grid maintenance, and emergency repair. In these safety-critical power-system environments, identity-continuous worker trajectories can provide a perception basis for on-site safety supervision, operational-procedure review, emergency coordination, and post-incident reconstruction. In practice, workers are frequently observed under helmet-induced facial occlusion, uniform work clothing, equipment clutter, and temporary disappearance behind cabinets, protection panels, inspection vehicles, tools, or other personnel. These conditions make conventional face recognition, body-only re-identification, and short-term tracking unstable. This study develops a helmet-aware cross-modal identity-consistent tracking framework as an edge-deployable perception module for intelligent power-system operation and maintenance. The framework extracts face, body, helmet, and motion representations using dedicated branches; estimates the reliability of each modality from visibility, detection confidence, and temporal consistency; fuses the available cues through reliability-normalized attention; and maintains a memory bank for re-identification after occlusion. The helmet branch explicitly decomposes helmet observations into illumination-normalized color, structural shape, and learned visual descriptors, which enables protective equipment to serve as a stable auxiliary identity cue instead of being treated only as an occluder. The architecture is trained end to end using a unified objective that combines identity classification, batch-hard triplet learning, helmet-attribute supervision, temporal consistency, and association losses. Experiments are conducted on the custom PGW-Track-85309 dataset, which contains real videos from substation and power-grid maintenance scenarios together with controlled occlusion protocols. Compared with face-only, body-only, DeepSORT, ByteTrack, OC-SORT, BoT-SORT, and TransReID-based tracking baselines, the proposed framework reaches 90.4% identity F1 (IDF1), 91.8% multiple object tracking accuracy (MOTA), and 77.6% higher-order tracking accuracy (HOTA) on the held-out test set, while reducing identity switches to 11. The model runs at 43.8 frames per second (FPS) on an RTX 3090 GPU and 31.7 FPS on an NVIDIA Jetson Orin NX edge device, meeting the real-time requirement of 30 FPS. The results indicate that domain-aware helmet modeling, reliability-aware fusion, and memory-based association jointly improve identity continuity under realistic power-grid occlusion. By preserving personnel identity through equipment-induced occlusion and reappearance, the framework strengthens site-level safety awareness and supplies reliable trajectory evidence for intelligent operation-and-maintenance decision support, while retaining human review for safety-critical decisions.

L. Meng, G. Ban, Jintong Ma et al. · 0 citations
Open access Aug 2026

Application of BERT-Free Pre-Training Model in Identifying Power Safety Risk Points

In intelligent power systems operating under increasingly complex electromagnetic environments, accurate identification of safety risk points is essential for ensuring reliable equipment operation and supporting electromagnetic compatibility assessment. Traditional rule-based methods suffer from limited semantic understanding and poor generalization, making them insufficient for processing complex operation and maintenance texts. To address this issue, this paper proposes a lightweight CNN-based pre-training model built on a BERT-Free architecture for efficient and accurate power safety risk identification. A professional dataset containing 58,000 power operation and maintenance texts is constructed, and a high-quality multi-label corpus covering four categories and 17 subcategories is established through dictionary-enhanced word segmentation and expert cross-annotation. The model is exported in ONNX format to facilitate flexible deployment in engineering applications. Experimental results demonstrate that the proposed model achieves an overall accuracy of 89.1% and an F1-score of 87.3%. Under class imbalance, the F1-score reaches 0.901 for majority classes and 0.784 for minority classes, exhibiting strong robustness. The average inference time per sample is only 8.3 ms, indicating high computational efficiency. These results demonstrate that the proposed model provides an effective and practical solution for intelligent power safety risk identification while offering valuable support for electromagnetic infrastructure monitoring and reliable operation in modern power systems.

S. W. Yu, Y. He, G. Ban et al. · 0 citations
Open access Aug 2026

Knowledge Distillation Method for Compressing Large Language Model of Power Risk Identification and Improving Deployment Efficiency

Experiments show that, after applying the proposed knowledge distillation method, the inference latency is reduced from 235 ms to a minimum of 26 ms, which is better than DistilBERT’s 35 ms, verifying the efficiency and practicality of the lightweight model in resource-constrained scenarios involving power-risk identification, electromagnetic sensing, and edge-based intelligent monitoring.

S.-W. Yu, Y.-M. He, G.-B. Ban et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.