Fault Detection Using Nonlinear Low-Dimensional Representation of Sensor Data

Recent advances in many enabling technologies such as sensing, computing and communication are instrumental in achieving this objective. Real-time health monitoring enables transitioning from traditional fixed schedule preventive maintenance to predictive maintenance, where decisions regarding maintenance are based on an objective assessment of the equipment health. The reduction in price of sensors has enabled widespread adoption of sensor technology for health monitoring. In 2004 the average cost of sensors was $1.30 and in the year 2020, it is expected to come down to $0.38 [1]. Industries such as mining, transportation, and aerospace are among the leaders in adoption of sensor-enabled predictive maintenance.