Answers are grounded in this paper and cite their sources.
To address frequent failures of high-pressure dust removal fans in steel plants under high-temperature and high-dust conditions and the difficulty of single-parameter threshold monitoring in detecting early degradation, an unsupervised online health assessment method based on Gaussian mixture model and Mahalanobis distance is proposed. Normal data construct health baselines for motor and fan-bearing subsystems, and weighted Mahalanobis distance is mapped to a health degree in [0,1] via an arctangent function. Experiments on 5.27 million field records achieved 100% recall, 90.5% precision, and 9.5% false discovery proportion on 38 abnormal events, outperforming traditional methods and isolation forest. Ablation experiments confirm the necessity of multi-component modeling and covariance-based measurement. Practical cases show early warning of progressive degradation and identification of multi-parameter coordinated anomalies. Inference time is only 0.0009 ms per record, requiring no fault labels, facilitating online deployment.
No takes yet. Share an insight, caveat, or question.
Xie et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: