2018/04/16 by Arvind Kumar Shekar, Cláudio Rebelo de Sá, Shekar, Arvind Kumar +6
Computer Science · Engineering · #Anomaly Detection Techniques and Applications #FOS: Computer and information sciences #Fault Detection and Control Systems #Industrial Vision Systems and Defect Detection #Machine Fault Diagnosis Techniques #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Sensor Technology and Measurement Systems
paper · pdf · doi:10.48550/arxiv.1804.05544
openalex publication_date 2018/04/16 · openalex created_date 2022/10/02 · openalex updated_date 2026/07/28
Predicting the health of components in complex dynamic systems such as an\nautomobile poses numerous challenges. The primary aim of such predictive\nsystems is to use the high-dimensional data acquired from different sensors and\npredict the state-of-health of a particular component, e.g., brake pad. The\nclassical approach involves selecting a smaller set of relevant sensor signals\nusing feature selection and using them to train a machine learning algorithm.\nHowever, this fails to address two prominent problems: (1) sensors are\nsusceptible to failure when exposed to extreme conditions over a long periods\nof time; (2) sensors are electrical devices that can be affected by noise or\nelectrical interference. Using the failed and noisy sensor signals as inputs\nlargely reduce the prediction accuracy. To tackle this problem, it is\nadvantageous to use the information from all sensor signals, so that the\nfailure of one sensor can be compensated by another. In this work, we propose\nan Artificial Neural Network (ANN) based framework to exploit the information\nfrom a large number of signals. Secondly, our framework introduces a data\naugmentation approach to perform accurate predictions in spite of noisy\nsignals. The plausibility of our framework is validated on real life industrial\napplication from Robert Bosch GmbH.\n