To ensure the stable operation of substations, staff members are frequently required to enter operational areas. However, due to the lack of proactive monitoring mechanisms, accidental intrusions into live zones often lead to electric shock incidents. To address this issue, this study proposes an intelligent method that integrates keypoint detection, ranging, and early warning into a unified framework. First, a Channel-aware Inception Depthwise Convolution (CIDW) feature extraction module is proposed to enhance the capability of small-object feature extraction. Second, a Context-Dynamic Attention (CDA) feature enhancement module is developed to improve the model’s ability to capture fine-grained details in critical regions. Third, a Context-Enhanced Mechanism (CEM) module is designed to improve the feature pyramid network, thereby strengthening the model’s perception of high-resolution features. To overcome the challenges that monocular cameras cannot directly obtain threedimensional information and that long-distance pixels exhibit nonlinear errors, an innovative Long-distance Error Correction (LEC) model is proposed. By combining an inverse coordinate transformation formula, the LEC model enables high-precision 3D ranging. Experimental results demonstrate that the proposed detector outperforms existing algorithms in terms of detection accuracy, speed, and lightweight design. The ranging accuracy reaches the millimeter level at short distances, with long-distance errors controlled within approximately 5 cm, making it an effective safety monitoring approach for the perception layer of intelligent substation IoT systems.

Current multi-modal sensor trackers mainly rely on visual cues for target tracking. However, in challenging scenarios, the visual information acquired by multi-modal sensors has limited descriptive ability when the target state undergoes significant changes, which may lead to unsatisfactory tracker performance. In this work, we propose TVTracker, a two-stage tracking framework. It leverages semantic information of the target state to enhance visual cues, enabling effective RGB-T tracking. The first stage is the generation of target text descriptions. Utilizing the Bootstrapping Language-Image Pre-training (BLIP) model, we generate target textual descriptions that match the images in the dataset. In the second stage, text-guided visual fusion is performed for target tracking. Target textual and visual features are extracted separately using the text and visual branches. Then the target textual features are integrated with the visual features to localize the target position and predict the target bounding box. In the text branch, we design the Target Text Adaptation Enhancement (TTAE) module to mitigate the interference of low-quality target textual features on visual features. In the visual branch, we develop the multi-modal visual information prompters, which include the Multi-modal Visual Shared Information Prompter (MVSIP) and the Multi-modal Visual Shared and Complementary Information Prompter (MVSCIP), to facilitate learning of multi-modal shared or complementary visual prompts. Experiments on the LasHeR, RGBT210, RGBT234, and VTUAV datasets demonstrate the effectiveness of TVTracker.

Base station networks are a crucial component of fifth-generation communication systems. Faced with increasing traffic demands and energy consumption, connecting base stations to microgrids is essential for optimizing resource management within green base station networks and reducing their energy consumption and environmental impact. In this situation, existing methods for renewable energy base station resource management lack flexibility and intelligent optimization for energy trading involving multiple base stations. Therefore, this paper proposes an energy trading method based on software-defined networking (SDN) and the nonlinear tangent perturbation-multiagent proximal policy optimization (NTP-MAPPO) algorithm that improves the economic efficiency and renewable energy utilization rate of multi-base-station microgrids. Specifically, we propose a reference scenario for energy trading within a multibase-station microgrid based on SDN, and then model it using game theory to account for energy sharing among different base station microgrids. We express this model as a Markov decision process and solve it using the improved NTP-MAPPO algorithm, which offers an enhanced exploratory performance. Numerical analysis is subsequently used to validate the effectiveness of the proposed method.

The precise location of insulators in infrared images is of great significance for insulator condition monitoring and fault diagnosis. Due to the characteristics of insulators themselves and the use of handheld infrared cameras, insulators usually appear in infrared images with different aspect ratios and main axis orientations. Therefore, it is very important and necessary to make full use of the prior knowledge of the insulator itself to accurately locate it. However, most of the existing methods use axial horizontal detection boxes to detect insulators, which cannot take into account the characteristics of the insulator well. When there are large overlapping areas of two horizontal detection boxes, the nonmaximum suppression algorithm may lead to missed detection of the object. To further improve the accuracy of the detection algorithm, this article makes full use of the prior features carried by the insulator itself, and optimizes faster region-based convolutional neural networks (R-CNN) from five aspects: rectangular box representation, feature extraction, candidate box generation, anchor design, and feature alignment. An oriented detection model for infrared images of insulators is constructed. Comparative experiments with a variety of mainstream detection methods were carried out on the constructed infrared dataset. The results show that the proposed method is superior to other models in detection accuracy. When the intersection and union ratio is 0.5, the average precision reaches 95.08%. In addition, it can also effectively predict the shape and angle information of insulators in complex scenes, laying a beneficial foundation for subsequent diagnosis automation tasks.

Moisture is one of the critical factors to determine the service life of transformers. The moisture inside the transformer oil-immersed insulation could be quantified with feature parameters. This article proposes and develops a genetic algorithm support vector machine (GA-SVM) model to carry out the moisture diagnosis. Present findings reveal that these feature parameters can be obtained by using frequency-domain spectroscopy. Therefore, a novel model for predicting the frequency-domain spectroscopy curves is first reported based on a small number of samples, which could be utilized to obtain the feature parameters database to develop GA-SVM. Then, the moisture diagnosis in the lab and field conditions is presented to verify its feasibility and accuracy. The novelty of this article is in an exploration of the reported model as an intelligent based moisture diagnosis tool for power transformers.