2019/10/12 by Daniel Jung, Jung, Daniel, D. Jung · 1 citation
Computer Science · Engineering · #Anomaly Detection Techniques and Applications #FOS: Computer and information sciences #FOS: Electrical engineering #Fault Detection and Control Systems #Machine Fault Diagnosis Techniques #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Signal Processing (eess.SP) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.1910.05626
openalex publication_date 2019/10/12 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Localization of unknown faults in industrial systems is a difficult task for\ndata-driven diagnosis methods. The classification performance of many machine\nlearning methods relies on the quality of training data. Unknown faults, for\nexample faults not represented in training data, can be detected using, for\nexample, anomaly classifiers. However, mapping these unknown faults to an\nactual location in the real system is a non-trivial problem. In model-based\ndiagnosis, physical-based models are used to create residuals that isolate\nfaults by mapping model equations to faulty system components. Developing\nsufficiently accurate physical-based models can be a time-consuming process.\nHybrid modeling methods combining physical-based methods and machine learning\nis one solution to design data-driven residuals for fault isolation. In this\nwork, a set of neural network-based residuals are designed by incorporating\nphysical insights about the system behavior in the residual model structure.\nThe residuals are trained using only fault-free data and a simulation case\nstudy shows that they can be used to perform fault isolation and localization\nof unknown faults in the system.\n