Processes for Condition Monitoring and Prognostics at the Sensor Node

Condition-based maintenance (CBM) has the potential to significantly reduce operating costs through improved equipment management and more informed scheduling of maintenance activities. Sensing hardware can monitor equipment health, apply diagnostic algorithms to evaluate an assets current state, and make projections for how the system will perform if damage is detected and tracked. For maintainers to benefit most from CBM and prognostics (CBM+) techniques, more processes need to be implemented at the sensor node where transient and steady-state operations are monitored. These computational processes themselves must be robust and reliable in identifying the type and extent of damage as it evolves, while also being resource and power efficient for deployment on low-power electronics platforms. Luna proposes to develop methods for processing time, frequency, and modal domain data for hydraulic subsystems that are representative of those on the Virginia class submarine. Physical and data driven statistical models will be used to characterize the systems normal operating state and subsequently used to generate failure estimates based on data from damage scenarios conducted in the laboratory. Algorithms will be written expressly for implementation on low-power microcontrollers, leveraging Lunas extensive experience in the design and fielding of embedded sensor nodes for condition monitoring.