Ruopeng Yan

Real-world time series forecasting is often challenging due to non-stationarity and distribution shifts, where the optimal forecasting model varies across different temporal regimes and horizons. In this work, we introduce a method called Adaptive Future-Guided Ensemble Learning (AFG-EL), a two-stage framework that performs drift-aware, sample-level routing over a heterogeneous model zoo. AFG-EL learns dynamic fusion weights from meta-features of the historical window and incorporates a future-guided training signal from a relative-future teacher or scorer, emphasizing learning on regime transitions and drift-sensitive segments. Crucially, the inference process remains strictly causal, requiring only historical data and extracted meta-features. We further use sparse routing with an entropy-based fallback mechanism to enhance stability when routing confidence is low. Our experiments on several commonly used forecasting datasets demonstrate that AFG-EL consistently outperforms strong single-model baselines, uniform averaging, and adaptive fusion baselines.

Bearing fault diagnosis is an important task for condition monitoring and predictive maintenance of rotating machinery. Nevertheless, many existing deep learning-based methods have difficulty in jointly modeling multi-scale fault characteristics, adaptively highlighting informative features, and maintaining robustness under noisy measurement conditions. To address these issues, this study presents MDCAD-Net, a multi-dilated convolution attention denoising network that integrates multi-scale temporal feature extraction, attention-based feature refinement, and explicit noise suppression within an end-to-end learning framework. Parallel dilated convolutions with different dilation rates are employed to capture short-duration transient impulses as well as long-range periodic patterns in vibration signals. Channel-wise feature recalibration using squeeze-and-excitation networks and spatial-temporal attention via a convolutional block attention module are combined to enhance informative representations. In addition, a denoising block with gated attention and residual connections is introduced to reduce noise interference while retaining fault-related signal components. Experiments conducted on the Case Western Reserve University bearing dataset show that the proposed method achieves a classification accuracy of 98.93% and yields competitive performance compared with several commonly used deep learning models. Ablation studies and feature visualization results further illustrate the contributions of the individual components and the separability of the learned feature representations under noisy conditions. The results indicate the potential of the proposed framework for practical bearing fault diagnosis under noisy operating conditions.

Edible oil plays an irreplaceable role in the diet worldwide. It not only grants unique flavors to the food but also provides the necessary nutrition we need for daily life. Therefore, the quality of edible oil directly affects our health conditions, and public health issues caused by unqualified edible oil occur every year, especially in low-income populations. Unfortunately, currently available edible oil detection methods usually involve bulky equipment and various reagents which are cost-intensive and not friendly to undeveloped areas. In this study, we design and fabricate a disposable sensor for edible oil detection with high sensitivity by measuring the refractive index (RI) of the edible oil using an arched optofluidic waveguide. The experimental results show that our sensor can measure the RI of the edible oil with a high sensitivity of 16.175 mW/RIU and a low noise equivalent detection limit (NEDL) of 4.099×10 ^-5 RIU simultaneously in a dynamic range from 1.45839 to 1.47399 RIU. In addition, our minia turized sensor can rapidly detect a sample within only 30 seconds, providing a highly efficient and convenient solution for edible oil analysis in various settings. The remarkable RI sensing capability enables the sensor to distinguish different types of commercial edible oil. Moreover, our sensor can accurately monitor the RI change of the edible oil as the frying time increases, which is potentially employed for nutrition/toxicological evaluation. We believe our sensor can provide a promising solution for cost-effective and high-sensitivity edible oil detection in underdeveloped areas and hence contribute to human health.

We optimized the microfluidic chip of imaging flow cytometry by setting horizontal connections; By OTS microscopy, cells can be imaged at 40 m s⁻¹in our device.

Refractive index (RI) sensing plays an important role in analytical chemistry, medical diagnosis, and environmental monitoring. The optofluidic technique is considered to be an ideal tool for RI sensor configuration for its high integration, high sensitivity, and low cost. However, it remains challenging to achieve RI measurement in real time with high sensitivity and low detection limit (DL) simultaneously. In this work, we design and fabricate a RI sensor with an arched optofluidic waveguide by monitoring the power loss of the light passing through the waveguide, which is sandwiched by the air-cladding and the liquid-cladding under test, we achieve RI detection of the sample in real time and with high sensitivity. Furthermore, both numerical simulation and experimental investigation show that our RI sensor can be designed with different geometric parameters to cover multiple RI ranges with high sensitivities for different applications. Experimental results illustrate that our sensor is capable to achieve a superior sensitivity better than −19.2 mW/RIU and a detection limit of 5.21×10⁻⁸ RIU in a wide linear dynamic range from 1.333 to 1.392, providing a promising solution for real-time and high-sensitivity RI sensing.