Jun Xie

Deep learning is extensively used in fault diagnosis. Actual implementations frequently encounter the issue of limited labeled data. This paper proposes a cross-domain self-supervised contrastive learning method with multi-scale feature fusion (CD-MSSCL) for bearing fault diagnosis under limited samples. The method first employs a specialized time-domain data augmentation strategy to capture the complexity of industrial vibration signals and enhance model generalization. A designed encoder backbone network combines multi-scale features with a SENet attention mechanism using pyramid fusion. The network performs self-supervised pre-training on unlabeled samples to effectively capture multi-frequency fault features critical for bearing fault diagnosis. Limited labeled samples then fine-tune the model to transfer pre-trained features to specific tasks. Experimental results show CD-MSSCL outperforms traditional deep learning and current contrastive learning methods in accuracy and domain adaptation under limited labels. The approach significantly reduces data collection and labeling costs through effective unsupervised knowledge extraction and transfer.

This study aims to improve the performance of auditory brain-computer interfaces (BCIs) by developing two-target and three-target paradigms based on steady-state motion auditory evoked potential (SSMAEP) using low-frequency stimuli in a spatial audio environment. SSMAEP is elicited by auditory stimuli exhibited by periodic and discrete changes in auditory spatial position. Methods: We designed a periodic auditory motion stimulation paradigm to evoke SSMAEP. Two-target and three-target SSMAEP-BCIs were developed. For the two-target SSMAEP-BCI, two periodic auditory motion stimuli with different motion frequencies were located on the left (2 Hz) and right (1.6 Hz) sides of the head, respectively. For the three-target SSMAEP-BCI, three periodic auditory motion stimuli with different motion frequencies were located on the front (2 Hz), left (2.4 Hz) and right (1.6 Hz) sides of the head, respectively. Results: SSMAEP amplitudes were modulated by auditory selective attention. In the two-target BCI, the offline experiments showed a peak average information transfer rate (ITR) of 7.70 bits/min, while the online experiments had a mean accuracy of 82.83% and an ITR of 4.41 bits/min. The three-target BCI achieved a peak ITR of 12.04 bits/min offline, with an online mean accuracy of 80.45% and an ITR of 7.05 bits/min. Conclusion: The study confirms the feasibility and enhanced performance of spatial low-frequency SSMAEP-BCIs. Significance: This novel approach to SSMAEP-BCI offers a promising direction for enhancing auditory BCI performance, potentially improving user experience and application in complex environments.

Modern video games are increasingly aiming for more natural interactions and healthier gaming experiences. Haptic devices, in particular, can enhance these experiences by providing multimodal feedback and simulating a variety of body postures. However, limited attention has been paid to utilizing upper limb wearable haptic devices in video games. In. this study, we developed a flight video game that incorporates a wearable pneumatic haptic device. Our designed haptic feedback sleeve can deliver changes in both haptic forces and applied areas on the forearm. The proposed device consists of 40 airbag units made from two layers of TPU film, sealed by heat. To verify its performance, we conducted finite element simulations and experiments to assess the output force, area, and linearity of the airbag units. Two haptic perception experiments were conducted to verify the distribution of this haptic feedback device. Finally, experimental validation combining the flight video game was conducted. The results showed that the distributed upper limb haptic feedback sleeve reduced the user's following angle error by 12.99% when the aircraft roll speed was 16 deg/s. This finding indicates an enhancement in limb motor control ability using the proposed haptic feedback sleeve.

The rapid advancement of brain-computer interface (BCI) technology has created interest in steady-state visual evoked potential (SSVEP)-based BCIs, which are valued for their high information transfer rates and ability to manage multiple targets. Nonetheless, the efficacy of SSVEP decoding is often constrained by the volume and duration of user calibration data, limiting its practical application. Generative adversarial networks (GANs) have shown promise in synthesizing SSVEP electroencephalogram (EEG) data. However, they face challenges, such as low signal-to-noise ratio and capturing both temporal and spatial features. To address the need for high-quality data generation, the Attention-Aided Classifier Wasserstein GAN (AAC-WGAN) is proposed, which is a novel GAN model that combines an Attention Mechanism and an Auxiliary Classifier to improve data quality and classification performance. Our experiments on the Direction and Dial datasets reveal that our model performs obviously better, particularly with a training set ratio of 25% synthetic data and 75% real data. It achieves a classification accuracy of 91.65% on the Direction dataset, which is a significant improvement over the baseline accuracy of 84.32% (p = 0.008) and outperforms other comparable models (p < 0.05 for all comparisons). On the Dial dataset, our model achieves a classification accuracy of 83.48%, outperforming both the baseline of 80.86% and other models. Additional analyses, such as t-SNE visualization and FFT, confirm our model’s effectiveness in feature representation and frequency domain characteristics, with significant improvements in both data quality and classification stability. This work advances the state of SSVEP generation and classification, resulting in significant improvements in data quality and accuracy. It also represents a significant step forward in the development of BCI technologies.