Bearings are an essential part of rotating machinery, but they frequently fail, particularly in situations requiring high-strength load bearing and fast rotation speed. This research suggests a unique strategy based on acoustic signals to solve the issues of low accuracy and limited resilience in the current bearing failure detection methods. A lightweight fault diagnosis network, MS-GhostNet V3, is designed to enhance the performance and efficiency of bearing fault detection. Firstly, the acoustic signals produced by rotating machinery are collected using a linear microphone array, capturing the spatial distribution and phase characteristics of the acoustic signal. To improve fault characteristic expressiveness, the time–frequency domain feature map of the acoustic signal is obtained using the short-time Fourier transform. Then the MS-GhostNet V3 network is used to realize bearing fault diagnosis. MS-GhostNet V3 is composed of multi-scale convolution fusion module (MCFM), ghost bottleneck module (GBNM) and classifier. To obtain multi-scale information, the multi-scale convolution parallel structure creates the MCFM. The GBNM, integrating the ghost module with the spatial and channel synergistic attention module, facilitates a more comprehensive representation of fault characteristics across both spatial and channel dimensions. Consequently, this enhances defect identification precision. According to experimental results, the suggested approach yields accuracies of 96.19% and 96.88% on both private and publicly available datasets. These outcomes satisfy the requirements for real-time defect detection and lightweight operation. And compared to classical methods, it exhibits superior fault diagnosis performance, with enhanced accuracy and robustness under noise interference. The code is available at: https://github.com/xgli411/MS-GhostNet-V3.

Low-light images often suffer from uneven illumination, resulting in reduced brightness, low contrast, and increased noise interference. However, existing enhancement methods frequently lead to over- or under-enhancement, frequency domain distortion, or insufficient noise suppression, which adversely affect both visual quality and the performance of downstream vision tasks. To address these issues, we propose AASFNet, a novel low-light image enhancement network that integrates spatial and frequency domain features through amplitude enhancement and dual-domain fusion. The framework consists of a Frequency-domain Enhancement Network (FreqEnhanceNet) with an illumination-guided Amplitude Module (AmpModule) for adaptive brightness adjustment and a Spatial-Frequency Fusion Network (SpatFreqFusionNet) incorporating a Dual-Domain Fusion Module (DDFModule) for noise suppression and illumination correction via multi-scale feature interaction. This design enables the network to simultaneously adjust global illumination, restore local details, and suppress noise. Additionally, a joint spatial-frequency loss function is introduced, including an illumination-guided amplitude loss and a cosine phase loss, to enhance structural consistency and amplitude fidelity. Comprehensive experiments on the LOL-v2 Real and Synthetic datasets demonstrate that AASFNet achieves competitive or leading performance across multiple key metrics under the evaluated experimental settings, yielding the best or second-best PSNR, SSIM, and NIQE values among the compared methods. Moreover, when applied as a preprocessing module, it achieves improved detection accuracy across multiple low-light benchmarks, with mAP scores of 78.63% on ExDark, 78.2% on DarkFace, and 80.32% on LoLI-Street. These results suggest the potential applicability of the model in real-world scenarios such as autonomous driving and surveillance, although further validation under actual deployment conditions is still required.

Longitudinal Tear Detection of Conveyor Belt Based on Improved MFCC and DenseNet Network