An adaptive prognostic methodology for sensor-driven component replacement and spare parts ordering policies
To date, engineers have primarily focused on the problem of using sensor data to assess the current and future health status of critical components or systems, whereas logisticians have paid a great deal of attention to efficiently controlling the flow of parts and other resources to ensure task/mission readiness. In isolation, these tools have limited impact for two main reasons: (1) Component-specific sensor-based data streams do not capture the traditional reliability characteristics related to the component's population, i.e. reliability and degradation characteristics of other similar components. In addition, they have not been fully exploited in maintenance related operational and logistical decision strategies. (2) Maintenance operational and logistical models generally assume failure to be a random process. This work addresses these challenges by developing an adaptive degradation-based prognostic framework for estimating statistical distribution of the remaining useful life. The distributions are revised and updated using real-time health monitoring information. These dynamically evolving remaining useful life distributions (RULDs) are integrated with high-level replacement and logistics decision models to enable "sense and respond" adaptive framework for component replacement and spare parts ordering, which is driven by real-time prognostic information.