Hui Wang

Detecting transient “click” sounds during connector insertion is pivotal for automotive assembly quality but remains intractable due to high-intensity, non-stationary industrial noise. This paper introduces a physics-aware generative demasking framework that integrates acoustic spatial priors with conditional diffusion modeling. We propose the spatially conditioned diffusion probabilistic model (SC-DPM), where an ambient reference signal acts as a physical constraint to steer the reverse diffusion process. By exploiting the spatial decay of insertion sounds, this mechanism effectively disentangles the target transient from the background noise manifold, reconstructing high-fidelity spectro-temporal features. Discriminative temporal patterns are extracted using causal random convolutional kernels with causal dilations and local proportion of positive values (LPPV) pooling. Experiments on real-world datasets demonstrate 93.3% accuracy. The proposed “restore-then-classify” paradigm significantly enhances robustness against acoustic variability, establishing a scalable methodology for precise industrial monitoring under extreme noise conditions.

Cardiovascular disease has become the leading cause of death worldwide, underscoring the urgent need for widespread cardiac monitoring, while the Electrocardiogram (ECG) remains the diagnostic gold standard, the complexity of its acquisition limits its long-term feasibility. In contrast, Photoplethysmography (PPG), ubiquitous in wearable devices, is increasingly adopted due to its accessibility. However, synthesizing ECG from PPG poses an intrinsically ill-posed inverse problem. Existing purely data-driven paradigms often neglect underlying biophysical mechanisms, resulting in a lack of physical constraints and interpretability, which renders them prone to generating non-physiological hallucinations. To address this, we propose PhysDiff-LBM, a novel physics-aware framework that incorporates Lattice Boltzmann hemodynamic constraints into a conditional diffusion model. Employing a dual-stream architecture, our framework captures high-frequency morphological details via a cross-attention-guided diffusion model with region-wise adaptability. Synergistically, we physically regularize the ECG synthesis by leveraging the mesoscopic streaming and collision operators of LBM. By forcing the synthesized waveform gradients to evolve consistently with hemodynamic momentum, this mechanism constrains the model to strictly adhere to the fluid dynamic conservation laws governing pulse wave propagation. Experimental results demonstrate that our method achieves superior signal fidelity and exhibits significant advantages in downstream clinical applications.