Pressure transmitters are widely used in the process industry for pressure measurement. The sensing line, a core component of the pressure sensor in the pressure transmitter, significantly impacts the accuracy of the pressure transmitter’s output. The reliability of pressure transmitters is critical in the nuclear power industry. Blockage is recognized as a common failure in pressure sensing lines; therefore, a novel detection method based on Trend Features in Time–Frequency domain characteristics (TFTF) is proposed in this paper. The dataset of pressure transmitters comprises both fault and normal data. This method innovatively integrates multi-scale time series decomposition algorithms with time-domain and frequency-domain feature extraction techniques. Initially, this dataset is decomposed into multi-scale time series to mitigate periodic component interference in diagnosis. Subsequently, via the sliding window algorithm, both the time-domain features and frequency-domain features of the trend components are extracted, and finally, the XGBoost algorithm is used to detect faults. The experimental results demonstrate that the proposed TFTF algorithm achieves superior fault detection accuracy for diagnosing sensing line blockage faults compared with traditional machine learning classification algorithms.
In medical time series disease diagnosis, two key challenges are identified. First, the high annotation cost of medical data leads to overfitting in models trained on label-limited, single-center datasets. To address this, we propose incorporating external data from related tasks and leveraging AE-GAN to extract prior knowledge, providing valuable references for downstream tasks. Second, many existing studies employ contrastive learning to derive more generalized medical sequence representations for diagnostic tasks, usually relying on manually designed diverse positive and negative sample pairs. However, these approaches are complex, lack generalizability, and fail to adaptively capture disease-specific features across different conditions. To overcome this, we introduce LMCF (Learnable Multi-views Contrastive Framework), a framework that integrates a multi-head attention mechanism and adaptively learns representations from different views through inter-view and intra-view contrastive learning strategies. Additionally, the pre-trained AE-GAN is used to reconstruct discrepancies in the target data as disease probabilities, which are then integrated into the contrastive learning process. Experiments on three target datasets demonstrate that our method consistently outperforms other seven baselines, highlighting its significant impact on healthcare applications such as the diagnosis of myocardial infarction, Alzheimer's disease, and Parkinson's disease. We release the source code at xxxxx.
Vision Language Models (VLMs) face challenges in effectively coordinating diverse attention mechanisms for cross-modal embedding learning, leading to mismatched attention and suboptimal performance. We propose Consistent Cross-layer Regional Alignment (CCRA), which introduces Layer-Patch-wise Cross Attention (LPWCA) to capture fine-grained regional-semantic correlations by jointly weighting patch and layer-wise embedding, and Progressive Attention Integration (PAI) that systematically coordinates LPWCA, layer-wise, and patch-wise attention mechanisms in sequence. This progressive design ensures consistency from semantic to regional levels while preventing attention drift and maximizing individual attention benefits. Experimental results on ten diverse vision-language benchmarks demonstrate that our CCRA-enhanced LLaVA-v1.5-7B model achieves state-of-the-art performance, outperforming all baseline methods with only 3.55M additional parameters, while providing enhanced interpretability through more regionally focused and semantically aligned attention patterns.
In medical time series disease diagnosis, two key challenges are identified.First, the high annotation cost of medical data leads to overfitting in models trained on label-limited, single-center datasets. To address this, we propose incorporating external data from related tasks and leveraging AE-GAN to extract prior knowledge,providing valuable references for downstream tasks. Second, many existing studies employ contrastive learning to derive more generalized medical sequence representations for diagnostic tasks, usually relying on manually designed diverse positive and negative sample pairs.However, these approaches are complex, lack generalizability, and fail to adaptively capture disease-specific features across different conditions.To overcome this, we introduce LMCF (Learnable Multi-views Contrastive Framework), a framework that integrates a multi-head attention mechanism and adaptively learns representations from different views through inter-view and intra-view contrastive learning strategies.Additionally, the pre-trained AE-GAN is used to reconstruct discrepancies in the target data as disease probabilities, which are then integrated into the contrastive learning process.Experiments on three target datasets demonstrate that our method consistently outperforms seven other baselines, highlighting its significant impact on healthcare applications such as the diagnosis of myocardial infarction, Alzheimer's disease, and Parkinson's disease.
Electromagnetic coils are indispensable components for energy conversion and transformation in various systems across industries. However, electromagnetic coil insulation failure occurs frequently, which can lead to serious consequences. To facilitate predictive maintenance for industrial systems, it is essential to monitor insulation degradation prior to the formation of turn-to-turn shorts. This paper experimentally investigates coil insulation degradation from both macro and micro perspectives. At the macro level, an evaluation index based on a weighted linear combination of trend, monotonicity and robustness is proposed to construct a degradation-sensitive health indicator (DSHI) based on high-frequency electrical response parameters for precise insulation degradation monitoring. While at the micro level, a coil finite element analysis and twisted pair accelerated degradation test are conducted to obtain the actual turn-to-turn insulation status. The correlation analysis between macroscopic and microscopic effects of insulation degradation is used to verify the proposed DSHI-based method. Further, it helps to determine the threshold of DSHI. This breakthrough opens new possibilities for predictive maintenance for industrial equipment that incorporates coils.