Power grid marketing inspection relies on work order data to identify rule violations. However, manual labeling of audit results is costly and often infeasible at scale. This study proposes a self-training partial least squares (STPLS) model to address this challenge by leveraging both labeled and unlabeled data for semi-supervised modeling. The method iteratively augments labeled samples using high-confidence predictions, improving calibration accuracy under limited supervision. Experiments on over 10,000 real-world grid work orders demonstrate the model’s effectiveness in semantic audit topic alignment and time-series prediction of work order volumes. Additional validation on a spectroscopic benchmark dataset confirms the generalizability of the approach. The proposed method supports efficient and scalable intelligent inspection in the power grid domain.

The rapid growth of generative artificial intelligence has led to an explosion of multimodal data, intensifying the demand for adaptive cross-modal retrieval systems. However, persistent semantic and distributional gaps across modalities continue to hinder effective alignment. To alleviate these challenges without incurring prohibitive annotation costs, unsupervised cross-modal retrieval (UCMR) has emerged as an attractive alternative; nevertheless, it often struggles to establish reliable semantic correspondences due to the absence of supervision. To address this limitation, we propose MiRA (Multi-granularity Consensus Representation Alignment), an unsupervised framework that enhances semantic consistency and representation robustness via hierarchical, noise-aware learning. MiRA progressively refines multi-granular features and aligns cross-modal representations through consensus-driven optimization. Specifically, it comprises two key components: (1) Progressive Agreement Clustering (PAC), which improves pseudo-label reliability by enforcing cross-layer consistency and reducing uncertainty from single-granularity representations; and (2) Cross-modal Robust Association (CRA), which leverages the refined pseudo-labels to guide representation learning under a hybrid contrastive-consistency objective, promoting discriminative alignment while suppressing noisy associations. Extensive experiments on four benchmark datasets demonstrate that MiRA consistently surpasses nine state-of-the-art UCMR methods, confirming its effectiveness in capturing hierarchical semantics and achieving robust cross-modal alignment without supervision.

Floating photovoltaic (FPV) systems experience continuous attitude variations under wave excitation, and accurate laboratory reproduction of such motion is essential for evaluating electrical and mechanical performance. However, push-rod actuators typically used in FPV simulators suffer from asymmetric dead-zone voltages during direction reversal, causing stagnation, phase lag, and peak clipping when controlled with conventional PID. This letter identifies this overlooked limitation and proposes a lightweight nonlinear PID controller with direction-dependent bias compensation that forces the PWM output to instantaneously surpass the dead-zone thresholds. Implemented on an STM32F407, the method eliminates reversal-induced pauses and enables smooth sinusoidal tracking. Experiments show that the proposed controller reduces phase lag from 0.176 s to 0.061 s (65.3%) and the sum of squared errors from 12.75 to 1.37 (89.3%), representing a substantial improvement over linear PID. The approach is computationally inexpensive and broadly applicable to actuator systems exhibiting dead-zone characteristics.

Fault injection attacks pose significant threats to cryptographic systems, including SM9, which is based on bilinear pairings. Despite the widespread use of SM9, research on its vulnerability to such attacks remains limited. This letter introduces an innovative fault injection strategy targeting the SM9 signature process. By introducing a single-bit fault into the X-coordinate of the private key and analyzing the resulting changes, we construct an attack model using differential analysis. Based on this model, we derive a quadratic equation and solve it in a finite field, successfully extracting the private key information. Experimental results validate the feasibility of this approach, revealing a critical vulnerability in SM9's design. Our findings offer valuable insights into enhancing the security of SM9 and provide a foundation for developing effective defense strategies.

Currently, passive detection of airborne targets mainly centers on radiometer imaging with real apertures and pointwise sampling imaging based on micro-element grid division. The drawback of this imaging approach is the contradiction between temperature sensitivity and scanning rate. The faster the scanning rate is, the shorter the dwell time of the antenna beam at each position, and accordingly, the system temperature sensitivity will decline. Furthermore, the enlargement of the antenna aperture is constrained by many aspects. To further enhance the resolution of passive detection of airborne targets, this paper conducts a simulation study on the detection effect of synthetic aperture radiometers for airborne targets.

This paper describes the topology and operation of a new family of asynchronous frequency dividers called Fractional Digital Regenerative Frequency Dividers (FDRFD). The operation is explained using a digital approach in the time domain and compared to the closest analog equivalent. The fractional behavior is intrinsic to the topology without the need for an external modulator. Integer ratios are also easily obtained and can serve as multimodulus dividers when required. Starting with the elementary version of this kind of divider, a generalization is made with repeated cell patterns in the closed loop for finer control of the fractional ratio. Some measurements on an FPGA-based implementation are provided to validate the theoretical study.

Physical-layer specific emitter identification (SEI) faces significant misclassification challenges of unknown devices, especially in closed-set environments and the scarcity of available labeled data. To address these limitations, we introduce a Few-Spot Self-Supervised Adversarial Augmentation Specific Emitter Identification (FS-SA2SEI) framework that integrates adversarial augmentation, and open-set recognition (OSR) mechanisms. FS-SA2SEI enhances spectral and multi-resolution feature extraction while enabling robust classification of known emitters and rejection of unknown ones. Evaluated on Wi-Fi and publicly available datasets (ADS-B, FIT/CorteXlab), the framework achieves a closed-set accuracy of 89.9%, surpassing prior benchmarks by >5%, and an open-set detection rate (OSDR) of 64.0% at a worst-case openness of 0.6. These results highlight FS-SA2SEI’s potential as a scalable solution for real-time SEI in resource-constrained IoT systems, where adaptability to dynamic environments and unseen devices is critical.

Recently, the rotating target detector has been widely used in remote sensing images. However, the existing methods often use a large number of preset rotating anchors to cover the target, ignoring the interference caused by the shape change of target with high aspect ratio on network training, which is mainly reflected in the following two points: 1) missing high-quality positive samples containing slender targets with critical location information; 2) the gradient with the sharply changing causes training instability. In order to meet these challenges, we have designed a sample assignment strategy that can adapt to targets with different aspect ratios, and a training strategy that can more stably and accurately regression the bounding box with high aspect ratio. Specifically, first of all, the designed Shape Adaptive Label Assignment strategy introduces a weight function based on the IoU. Secondly, Gradient Equalization Regression Loss function is proposed to effectively alleviate the gradient instability of large aspect ratio targets during regression and make the model have better convergence. A series of experiments on DOTA and HRSC2016 datasets have confirmed the effectiveness of the proposed strategy.

Versatile Video Coding (VVC) introduces advanced techniques such as sophisticated intraprediction mechanisms that significantly improve compression efficiency. However, the increased complexity in intraprediction presents new challenges in computational load and power consumption, essential considerations for efficient video coding. This paper proposes a novel framework that combines discrete wavelet transforms (DWT) and histogram analysis to optimise the prediction directions in VVC. Through selective reduction of prediction directions based on dominant texture orientation and minimising residual energy, the framework achieves enhanced computational efficiency while maintaining video quality. Experimental results confirm its effectiveness, showing minimum rate distortion performance and superior encoding efficiency compared to other methods in the literature.

In this letter, a VAQ-based DAC switching scheme is proposed to improve the power efficiency of SAR ADCs. The input signals are sampled onto the bottom-plates of the most significant bit (MSB) capacitors, thereby eliminating the reset energy. The reference voltage VCM rath than VREF is switched during the third-bit cycle, thereby significantly reducing the power consumption. Additionally, an energy-efficient one-sided switching technique is employed from the fourth-bit cycle. This proposed switching scheme achieves a 99.51% reduction in switching energy over the classic scheme.

Two neural networks with two features (SNR and mutual information) are applied to link adaptation that is targeted for meeting the WLAN standard. The usefulness of neural-network-based link adaptation is validated for the WLAN and also the performance (throughput and latency) is analysed to obtain some practical gain of link adaptation based on neural networks over the traditional approach.

In this paper, a novel reconfigurable angular cone antenna that employs a rotatable fan ground plane for reconfiguration is introduced. By regulating the number and orientation of external grounds, the radiation range is progressively broadened, enabling the antenna to exhibit diverse radiation modes. Experimental results demonstrate that the antenna can successfully switch between three modes and validate seven distinct beam directions. These beam directions include: (1) a single ground plane at 45°, 135°, 225°, and 315°; (2) two ground planes at ±45°; and (3) four ground planes creating an omnidirectional radiation pattern. Notably, the antenna boasts a broad bandwidth of 60.66% (covering 1.2-4.9 GHz; VSWR < 2.0) which cover 2G/3G/4G/5G communication bands, while maintaining an antenna efficiency exceeding 90%, marking a significant performance enhancement.

We reported a graphene field-effect transistor (GFET) based on CVD Ge-based graphene and primarily investigated the low-temperature electrical characteristics. The self-alignment technique was used to fabricate GFET to reduce parasitic effects and improve transconductance and cut-off frequency. To further explore the electrical properties, we studied the direct current and radio frequency characteristics of the GFET over a temperature range from 4.2 K to 300 K, considering the temperature-dependent resistivity of intrinsic Ge. The DC characteristic of the GFET for 110 nm gate length, particularly the transconductance performance, exhibits a tiny variation of only ~ 5% across this temperature range. However, the cut-off frequency experiences a considerable increase, improving several tens of times when the temperature decreases to 4.2 K, with a maximum value of 3.49 GHz. This work illustrates a meaningful advancement in applying GFETs in the lowtemperature, high-frequency domain.

A novel enlarging fractional bandwidth (FBW) technique for acoustic-wave-lumped-element resonator (AWLR) -based bandpass filters is present. The new technique is based on a series of matching inductors that replace the parallel or series acoustic wave (AW) resonators of AW filters to beak the bandwidth constraint of the ladder-type structure. In addition, the transmission response analysis of the proposed AWLR-based filter is provided. A prototype of the AWLR-based filter is fabricated and tested to validate the proposed FBW widening technique. The measured insertion loss, FBW, and out-of-band rejection are 1.5dB, 1.63kt2 (kt2 is the electromechanical coupling coefficient of the AWR), and 38dB, respectively.

The fusion of renewable energy into power grid planning poses significant challenges. In this investigation, we assess planning strategies across eight key indicators encompassing economic, reliability, environmental, and adaptability considerations. Leveraging the Analytical Hierarchy Process (AHP) alongside the Entropy Weight Method, we ascertain both subjective and objective weights, subsequently synthesizing comprehensive weights with parsimonious discriminative data. Scores for planning strategies are then derived utilizing these weights. To illustrate and validate the efficacy of the proposed methodology, we apply it to four schemes within the IEEE-6 node system, incorporating wind and photovoltaic power plants.

Pooling operations, essential for neural networks (NNs), reduce feature map dimensions while preserving key features and enhancing spatial invariance. Traditional pooling methods often miss the feature maps’ alternating current (AC) components. This study introduces a novel global pooling technique utilizing spectral self-attention, leveraging the discrete cosine transform (DCT) for spectral analysis and a self-attention mechanism for assessing frequency component significance. This approach allows for efficient feature synthesis through weighted averaging, significantly boosting TOP-1 accuracy with minimal parameter increase, outperforming existing models.

Pseudo-spark switches have been successfully used in a variety of pulsed-power applications, including the European Large Hadron Collider, the Z-hoop fusion device, and the Dense Plasma Focus device. For such applications, it is important to reduce the trigger energy of the switch to save cost and improve stability. In this paper, a laser-triggered pseudo-spark switch is designed, which is triggered by a laser with a wavelength of 532 nm. Increasing the laser energy reduces the delay and jitter of the switch, and the laser energy threshold of 1.5 mJ for achieving switch stability can make the jitter of the switch less than 1 ns; in addition, increasing the hydrogen pressure inside the tube reduces the delay and jitter of the switch; and when the triggering energy is sufficiently large, the switch delay and jitter do not change with the switch when the anode voltage is changed, and the switch delay and jitter do not change with the switch. When the trigger energy is large enough, changing the anode voltage, the delay and jitter of the switch do not change with the switching voltage.

A robust interacting multiple model approach is proposed to address the problem of accuracy and non-Gaussian measurement noise in maneuvering target tracking. Firstly, the interacting multiple model effectively improves the accuracy of maneuvering target tracking. Secondly, the multiple fading factors are introduced into the prediction covariance matrix for adjusting the gain matrix in real-time to enhance the accuracy caused by model mismatch and improve the ability of state transitions. Finally, the maximum correntropy criterion effectively promotes the robustness to outliers. The effectiveness of the proposed approach has been verified through simulation.