Risk management and control of on-site power operations is an important link in the power production process. Traditional methods that rely on manual execution of safety rules lack proactive prevention and control capabilities, and insufficient efficiency leads to an increasing frequency of safety issues in power production. Therefore, it is urgent to adopt intelligent methods to improve the level of safety risk management and control. In response to this issue, this paper proposes a multimodal causal knowledge graph based auxiliary decision-making method for safety risk management in power field operations. Based on the characteristics and requirements of safety risk management for on-site power operations, first obtain multimodal data related to safety risk management for on-site power operations, including video, photo, and text information collected by end devices, as well as historical archive data in various system databases. Then, based on historical risk management rules, the causal relationships of the core elements of on-site work safety issues are sorted out, and the data involved in the multiple types of antecedent influencing factors of historical faults and safety accidents are taken as the causes, and the faults and safety issues are taken as the results, forming a complete causal knowledge sample and constructing a causal effect estimation learning model. This model can output the safety risks of on-site power operations, forming a causal knowledge graph to provide auxiliary support for early warning and pre control decision-making of safety risks in on-site power operations.

With the rapid improvement of the performance of large models, pre-training a large-scale model and fine-tuning on downstream has become a trend. However, previous knowledge distillation methods mainly concentrates on a specific downstream task. This paper proposes a novel distillation framework Mutual Distillation for Masked Autoencoder Model(MMIM) which transfers knowledge from large-scale pre-trained visual models to provide smaller backbones for downstream tasks, thereby saving time. In our model MMIM, to maintain the advantages of mutual learning, two different student models learn collaboratively to boost both feature representation capabilities simultaneously. At the same time, in order to utilize powerful feature representation capability of pre-trained teacher model, we use the intermediate features of the teacher model to supervise the learning of the student models. Experiments on ImageNet demonstrates the compressed model obtained by our framework possesses powerful feature representation comparable to teacher model. In addition, MMIM also outperforms state-of-the-art knowledge distillation methods which directly distills fine-tuned models. Keywords: Masked Image Modeling; pre-trained Model; Knowledge Distillation; Vision Transformers

Safety control is one of the most important tasks for power grid operation. However, safety control and risk warning models with high accuracy always have multitudinous parameters and need scaled resources for calculation. So as to sort out this problem, we proposed a novel and adaptive model lightweight method based on knowledge distillation. In our method, the intermediate features of the teacher network is utilized to guide the learning of the student network and a regressor is introduced behind the selected middle layer of teacher network to realize dimension match. The prediction confidence of teacher network is mapped to the feature space for the regressor initialization. Besides, the prediction confidence is used to construct the distillation temperature to improve the adaptability of information transfer. Experiments on ImageNet, CIFAR 100 and Grid-specific datasets demonstrate the lightweight model obtained by our method possesses powerful and adaptive feature extraction and representation ability.

Ensuring safety compliance is paramount in substation operations. However, worker re-identification (Re-ID) remains highly challenging due to severe occlusions, uniform appearance similarity, and substantial illumination variations across shifts and environments. Moreover, the escalating cost of manual identity annotation in large-scale, multi-site surveillance systems necessitates annotation-free approaches for practical deployment. In this paper, we propose AdaInCV (Adaptive Intra-Class Variation Contrastive Learning), an unsupervised Re-ID framework tailored for substation worker safety monitoring. The proposed method quantitatively evaluates the model’s learning capacity for each pseudo-cluster by measuring intra-class feature variation after DBSCAN clustering, and adaptively selects training samples with appropriate difficulty throughout the learning process. To this end, two novel strategies are introduced. Adaptive Sample Mining (AdaSaM) progressively constructs reliable identity clusters while dynamically updating the memory dictionary. Adaptive Outlier Filtering (AdaOF) further exploits informative outlier samples—primarily caused by heavy occlusion or extreme illumination—as hard negatives to enhance contrastive representation learning. Extensive experiments on two widely used Re-ID benchmarks (Market-1501 and MSMT17), as well as an in-house Substation Worker Re-ID (SWRID) dataset, demonstrate that AdaInCV achieves state-of-the-art performance with significantly faster convergence than existing methods, establishing a practical foundation for intelligent safety supervision in power grid operations.

Real-time estimation of the safety distance between power operation workers and energized equipment is essential for preventing electric shock, arc flash injury, and equipment accidents during live or near-energized operations. However, large-scale deployment remains difficult because conventional binocular ranging systems require accurate offline calibration, while monocular vision suffers from metric-scale ambiguity. This paper proposes a weakly calibrated multi-camera vision framework for safety distance estimation in power operation scenes. First, a cross-camera tracking strategy combining mask-guided ReID features, Kalman filtering, spatio-temporal constraints, and Hungarian matching is developed to maintain globally consistent worker identities. Second, a region-wise self-supervised depth estimation model is introduced: photometric consistency is imposed within worker regions, while cross-view feature reprojection is imposed within rigid equipment regions. Intrinsic parameter regularization and scene-inherent dimensional priors are then used to recover metric scale without placing additional calibration targets. Third, the YOLOv8-seg backbone is enhanced with a CNN-Transformer C2f_CT module to improve pixel-level segmentation of workers, energized equipment, and structural components under cluttered backgrounds. Experiments on three representative power operation scenarios show that the proposed method achieves 92.3% MOTA, 94.2% Rank-1 accuracy, 82.0% mAP, 8.7% relative distance error, and 22 FPS end-to-end speed. When using five-frame smoothing and uncertainty threshold, the proposed method can achieve 98.7% warning precision, 0.9% missed alarm rate and 1.3% false alarm rate. The results indicate that the proposed framework provides a practical balance between deployment cost, geometric accuracy, and real-time warning capability.