Risk assessment of sensor failures in a condition monitoring process; degradation-based failure probability determination

Abstract

Condition monitoring of a system is in need of an efficient sensor network for detecting system faults in early stages. Sensor is a silent-failure component as it is not monitored in the main system misleading in estimation of the system health state. Then, failure-causes of the system are not detected due to missed detection. A risk-based measure is proposed in this research to take into account the consequence of sensor missed-detection. All possible combinations are determined for failure-causes as components state vectors. The probability of missed-detection is estimated by applying occurrence probability of failure causes and degradation-based failure probability of sensors. First, a sensor characteristic is selected through which the degradation process is affected. Then a degradation model is developed to calculate the sensor time to failure. The consequences of missed-detection are also corresponding quantifiable potential losses through both failure costs and maintenance expenditure. Finally, all feasible sensors placement scenarios are compared due to proposed risk measure. As a case study, sensor network of steam turbine condition monitoring is selected. Various sensor placement scenarios of steam turbine are prioritized based on the risk index and results are discussed.