Remote sensing (RS) image scene classification is a challenging task due to its intraclass variety and the interclass similarity. Recently, many convolutional neural network (CNN)-based methods explore the network to handle this task. However, RS images usually have confusing background in addition to the relevant objects, and features only derived from the whole RS images cannot achieve satisfying results. To solve the problem, this letter proposed a method of utilizing the attention network to localize multiscale discriminative regions of the RS scene images and combining features learned from the localized regions by a classification network. Specifically, the classification network is composed of three subnetworks, which are trained by certain scaled regions separately. To learn more discriminative feature representations, feature fusion module is introduced to fuse the features of the three subnetworks in a more effective way. Experiments conducted on the AID and NWPU-RESISC45 data sets evaluate the effectiveness of the proposed method.

Developing an accurate partial discharge (PD) monitoring system for switchgear has attracted significant attention in recent times. Detecting and distinguishing PDs with a portable PD detector is challenging due to the inherent noise interference and the similarity among different PD signals in field conditions. In this study, we propose an innovative approach that combines knowledge distillation (KD) and multitask learning (MTL) to efficiently transfer valuable insights acquired by an advanced network to a more lightweight student network. Notably, a novel energy-adaptive sampling strategy is applied to frequency feature to represent acquired pulse waveforms efficiently. Furthermore, a novel spatial information distillation module is also proposed to enhance the knowledge transfer, thereby enhancing PD recognition efficiency and accuracy. Signals captured in substations undergo clustering through a $k$ -means model, effectively separating PDs from noise. Using the phase resolved PD (PRPD) patterns generated from clustered signals, we train our PD classification network. With the integration of KD and MTL modules, our PD classification model outperforms the baseline model in terms of both speed and accuracy. A comprehensive assessment of each proposed module was conducted through ablation studies in our experiments. Furthermore, our constructed PD recognition dataset was used to conduct a comparative analysis against other methodologies, showcasing the superior performance of our approach.

As a new emerging task, video corpus moment retrieval (VCMR) aims to find the video segments relevant to a given natural language query from a large number of untrimmed videos. It mainly includes two subtasks, finding the most relevant video based on the query text (video retrieval), and locating the segment most relevant to a given query in a video (moment localization). At the same time, since videos often contain rich multi-modal information such as audio, text, and images, how to align and interact with the multi-modal information of videos and the text information of natural language queries across modalities is the core issue of this task. This article proposes a Deformable Multigranularity Feature Fusion with Adversarial Training Network (DMFAT), first inputs the subtitle and frame multi-modal information of the video into our Multi-Scale Deformable Attention module and performs multi-granularity feature fusion through Deformable Attention respectively. Then, guided by the query, adaptive weights are generated to fuse the two multi-granularity modality features of the video. Finally, the cross-modal representation of the query and video features is obtained through a bidirectional attention module, and an adversarial contrastive learning objective is introduced to enhance more precise moment localization. Our model is evaluated on two representative video corpus moment retrieval benchmarks: TVR and DiDeMo. Extensive experiments have been conducted to demonstrate that our method outperforms existing work.

Ultra-high frequency (UHF) partial discharge (PD) detection is essential for routine maintenance of gas-insulated switchgear (GIS). Realizing low-noise detection with cost-effectiveness and ease of installation is promising but challenging. This paper presents a novel broadband noise-shaping network (BNSN) integrated with a printed monopole antenna (PMA) sensor to realize sufficient noise rejection and good impedance matching. Moreover, by analyzing the equivalent noise circuit and codesigning a high-speed preamplifier circuit with the proposed sensor, the detection sensitivity and anti-interference of the proposed sensor module are improved. The effectiveness and superiority of the proposed sensor are first verified in accordance with IEC 62478, followed by validations using a PD simulator and a real 24-kV GIS. The experimental results demonstrate that the proposed sensor module has the capability to identify different PD types even in noisy environments.

Fast and accurate recognition of partial discharge (PD) patterns is essential to prevent insulation failure-related outages of industrial equipment. The ultrahigh-frequency (UHF) methods have been widely used for PD diagnosis due to their good versatility and noncontact capabilities. However, in harsh environments with significant noise and electromagnetic interference (EMI), existing UHF methods are difficult for acquiring high-quality PD signals and differentiating PD types. To overcome this issue, this article proposes an innovative method by combining a low-noise UHF sensing frontend and a wavelet scattering feature extraction network (WSN). The developed sensing frontend consists of a UHF antenna sensor and a conditioning circuit with bandpass configuration and stability compensation. The equivalent noise circuit of the frontend is modeled to analyze and optimize output noise. In addition, the constructed WSN with improved configuration directly derives low-variance features from noise-corrupted time series UHF PD signals. Subsequently, a low-complex and robust majority voting-based support vector machine (MVSVM) is trained to identify different PD types and EMI. In accordance with IEC 62478, experimental case studies validate the noise performance of the developed UHF frontend and demonstrate the effectiveness of the proposed WSN for PD classification. The proposed method achieves 91.3% accuracy on noisy datasets. Moreover, it surpasses comparison methods by 8.9% to 24.1% on insufficient datasets with significant noise and EMI.

Developing intelligent methods for partial discharge (PD) diagnosis, capable of handling various types of insulation defects in switchgear, has garnered significant attention in recent years. Certain PD signals exhibit similar characteristics, often leading to their confusion with noisy signals during data acquisition. To mitigate noise interference and enhance the precision of PD recognition, this article introduces a novel framework for separating PD signals from noise and acquiring discriminative features for identifying different types of PDs. Specifically, the proposed approach incorporates an adaptive frequency sampling strategy to extract effective and efficient features for the separation of PD signals and noise, followed by the clustering of the captured signals. Phase Resolved PD (PRPD) patterns are then generated for each clustered signal group, forming the PRPD pattern database. In order to identify the informative region within the PRPD patterns, we introduce spatial correlation attention and discriminative feature learning modules. These modules aim to reduce intraclass variance and increase interclass differences in the PRPD patterns. To evaluate the effectiveness of the proposed method in separating PD signals from noise and recognizing different PD patterns, we constructed a PD recognition dataset that encompasses noise as well as three types of PDs: 1) corona, 2) internal, and 3) surface. By conducting experiments and comparing the results with state-of-the-art methods, we demonstrate the performance of our method in achieving accurate PD recognition with a notable improvement of 1.9% on the constructed PD dataset.