Aug 2026· Energies· Vol 19, pp. 3700· 0 citations· 22 references
TL;DR
A lightweight Coordinate Attention ResNet network named ResNet–LCA is proposed based on the dual-modal time–frequency fusion of the Continuous Wavelet Transform and Short-Time Fourier Transform, which achieves coordinated optimization of classification accuracy, noise immunity and inference efficiency and can provide lightweight technical support for online intelligent power quality monitoring at the edge nodes of microgrids and islanded power systems.
Abstract
With the continuous penetration of renewable energy and power electronic equipment into modern power systems, the occurrence frequency of composite power quality disturbances has increased significantly. The accurate classification of various composite disturbances under strong noise remains a critical technical challenge. The existing single time–frequency transformation methods cannot simultaneously capture transient time-domain details and fine frequency-domain features of steady-state harmonics, while mainstream deep learning classification networks contain redundant parameters and introduce excessive computational overhead, failing to meet the real-time deployment requirements of power edge terminals. To address these limitations, a lightweight Coordinate Attention ResNet network named ResNet–LCA is proposed based on the dual-modal time–frequency fusion of the Continuous Wavelet Transform and Short-Time Fourier Transform. First, the two transforms are implemented separately to generate two groups of complementary time–frequency maps, which are concatenated along the channel dimension to fully extract the coupling features between the steady-state harmonics and the transient impulses. Second, a Haar wavelet subband mean aggregation module is designed for dimensionality reduction with negligible information loss. This module eliminates the channel redundancy introduced by the multimodal fusion and reduces the overall computational overhead at the input stage. Finally, a lightweight residual network integrated with Coordinate Attention is constructed, with Grouped Half-Convolution adopted to compress the model parameters. CA offsets the feature attenuation induced by the lightweight structural design and further improves the model’s noise immunity. A simulation verification was carried out on a simulated dataset covering 25 types of single and superimposed composite disturbances. At a signal-to-noise ratio of 20 dB, the proposed method achieved an average classification accuracy of 97.92%, with only 5.32 M total parameters and a single-sample GPU inference latency of 0.33 ms. Compared with standard ResNet-18 under 20 dB noisy conditions, the total parameter volume was reduced by 52.7%, the inference latency was shortened by 0.13 ms, and the classification accuracy was improved by 0.60 percentage points. The proposed method achieves coordinated optimization of classification accuracy, noise immunity and inference efficiency, and it can provide lightweight technical support for online intelligent power quality monitoring at the edge nodes of microgrids and islanded power systems.
A deep learning framework based on a Mamba-driven state-space model architecture for comprehensive PQ disturbance classification is proposed and results indicate that the proposed method is well-suited for real-time smart grid monitoring and intelligent protection systems.
Pintu Das, Chandan Jana, Sannistha Banarjee et al.· Engineering Research Express· 0 citations
Acoustic-vibration multimodal fusion technology has developed rapidly in the field of equipment fault diagnosis due to its ability to effectively suppress noise interference and enhance diagnostic reliability. However, existing acoustic-vibration multimodal fusion methods have not been adapted to the transient impact characteristics and structural features of circuit breakers, and still suffer from issues such as modal heterogeneity, feature redundancy, poor noise robustness, and insufficient real-time performance. To address these challenges, this article proposes an acoustic-vibration multimodal fusion and shared-private decoupled network method for high-voltage circuit breakers. For vibration signals, a wavelet synchronous compression transform (WSST) is employed to enhance time-frequency resolution and accurately capture transient impact characteristics. For acoustic signals, per-channel energy normalization-Mel spectrum is constructed to suppress steady-state background noise and enhance transient fault features. Based on the modal decoupling theory, a shared-private dual-branch encoder is designed to decouple cross-modal-shared information from single-modal unique features, and combined with the self-attention mechanism to achieve deep fusion of multimodal features. The experimental results show that the proposed method achieved a diagnostic accuracy of 98.02% on the test set. Compared with existing diagnostic methods, this method offers higher diagnostic accuracy, greater robustness, and better engineering practicality, providing a viable technical solution for intelligent online fault diagnosis of high-voltage circuit breakers.
Kai Zhang, Hongming Lu, Jinning Chen et al.· IEEE Sensors Journal· 0 citations
Power quality disturbances (PQDs) in microgrids pose significant challenges to stable and reliable operation, particularly in tourism-oriented island systems with highly variable and uncertain load patterns. This paper proposes an interpretable PQD identification framework based on time–frequency feature extraction. The method utilizes a short-time Fourier transform to capture the nonstationary characteristics of voltage signals and constructs a compact feature set integrating time-domain, frequency-domain, and time–frequency information for disturbance classification. A supervised learning model is employed to map the extracted features to disturbance categories, while interpretability is achieved through feature contribution analysis, enabling explicit linkage between model decisions and the physical characteristics of PQDs. The proposed approach is validated using a combination of synthetic datasets, simulation data derived from MATLAB/Simulink R2024a microgrid models, and experimentally measured signals from a hardware-based platform. Case study results demonstrate that the proposed framework achieves a high overall classification accuracy of 99.50% across multiple disturbance types, including voltage sag, voltage swell, harmonic distortion, voltage flicker, transient disturbances, and hybrid disturbances. The interpretability analysis further confirms that the identified features are physically consistent with the underlying disturbance mechanisms. Overall, the proposed framework provides an accurate, robust, and interpretable solution for PQD identification, offering practical value for real-time monitoring and intelligent operation of renewable-rich microgrids.
Dilixiati Hayireding, Jun Hang, Lei Yang· Electronics· 0 citations
The increasing proportion of wind power integration in the power systems and the dynamic interaction of power electronic equipment lead to frequent subsynchronous oscillation, which seriously threatens the safety and stability of the system. Therefore, it is urgent to develop a high-precision and robust identification method. Traditional standalone identification methods are vulnerable to wind speed fluctuations and noise, resulting in unsatisfactory accuracy and robustness. To accurately extract the oscillation parameters from the active power signal, this paper proposes a hybrid method for identifying subsynchronous oscillation in power systems. First, the active power signal is preprocessed using the wavelet threshold denoising strategy, which effectively filters out noise through multi-scale decomposition and signal reconstruction. Second, VMD is applied to the preprocessed signal to decompose it into intrinsic mode functions, thereby achieving effective separation of different oscillatory characteristics. Finally, the fast Fourier transform is used to perform spectral analysis on each IMF to accurately capture the dominant frequency and amplitude of each oscillating component. On the four-machine two-area system with DFIG, the verification is carried out under the wind speeds of 9 m/s, 10.5 m/s and 12 m/s and the no noise, 60 dB and 40 dB noise conditions. SSO is excited by inserting 20 Hz, 30.5 Hz or 40 Hz subsynchronous oscillation components into the external part of the mechanical power input of the generator. The results show that the relative error of frequency identification of the proposed method is less than 0.0426% under all test conditions. The relative error of amplitude identification is less than 1.4155%. Compared with different methods, the proposed method exhibits excellent performance under noise conditions and has robustness to wind change and noise interference.
Jinping Liang, Yi Zheng, Xiangde Mao· Electronics· 0 citations
To address the limitations of traditional Voltage-Current (VI) trajectory features in appliance load identification—such as the difficulty in distinguishing similar appliances, weakened amplitude information, and the absence of dynamic characteristics—this paper proposes a dual-stage cyclic training method for load identification that integrates multi-feature reconstruction with Particle Swarm Optimization (PSO). First, to overcome the high similarity of original VI trajectories, a PSO-based threshold optimization algorithm is designed to reconstruct VI trajectories through reflection operations and normalization, thereby enhancing the geometric morphological differences among similar appliances. Second, to supplement dynamic impedance information and energy level features, conductance-time trajectories and mean-square current color-block maps are introduced to characterize dynamic impedance variations and energy level differences, respectively. Finally, a three-channel classification network based on ResNet18 is constructed, where the reconstructed VI trajectories, conductance-time trajectories, and mean-square current color-block maps are fused via RGB channels as inputs, forming a closed-loop “threshold optimization—feature reconstruction—cyclic training” framework. Experimental results on the PLAID dataset demonstrate that the proposed method achieves an identification accuracy of 98.29% and a macro-averaged F1-score of 97.93%. Comparative experiments verify the effectiveness of the reconstructed VI trajectories, the complementarity of multi-feature fusion, and the superiority of the combinatorial optimization network, significantly improving the identification of multi-state and similar-condition appliances.
Yubo Wang, Shuai Zhang, Zhiyou Cheng· Italian National Conference...· 0 citations
Driven by global clean energy strategies, wind power develops rapidly. Bearings, core wind turbine transmission parts, govern system reliability and safety. Conventional diagnosis suffers three key practical limitations: single-sensor signals cannot fully characterize nonlinear composite faults; mainstream deep learning models act as opaque black boxes without clear diagnostic interpretability; highly coupled composite fault features cannot be separately extracted by existing algorithms. To address these challenges, this paper proposes an interpretable multi-sensor bearing nonlinear composite fault diagnosis method for wind power systems. Firstly, a multi-sensor dynamic frequency guided synchronous compressed wavelet transform is designed to precisely extract and unify multi-sensor non-stationary signal features via dynamic frequency matching, adaptive wavelet basis selection, and scale parameter optimization. Secondly, a dynamic calibration and feature enhancement network is constructed, including a dynamic dual-branch calibration fusion module for adaptive feature weighting and decoupling, and a wavelet attention feature enhancement network for sensitive feature enhancement and interpretability improvement. Finally, a Mahalanobis distance aware Krylov Transformer network is developed, integrating Mahalanobis distance to enhance early subtle fault sensitivity and an efficient global enhanced Krylov Transformer for deep feature modeling. Experiments across the three datasets yield average diagnostic accuracies of 98.80, 99.24, and 99.93%, respectively. Even under −4 dB noise interference, the proposed model retains an average accuracy above 90%. Component decoupling verification reveals that 96.0% of composite fault samples can be simultaneously identified via two independent fault channels, with the Pearson correlation coefficient between channel outputs as low as 0.18. Moreover, controlled sub-band masking tests show that masking the HH sub-band containing fault impulse information reduces the overall diagnostic accuracy from 98.45 to 76.89%, corresponding to a 21.56 percentage point drop, this sufficiently proves that the model’s inference relies heavily on high-frequency time–frequency features corresponding to fault impulses.
Sen Li, Xiao-qiang Zhao, Jie Cao et al.· Structural Health Monitoring· 0 citations