Combining Physics-Based Models with Signal-Based Methods for Machinery Diagnostics in Time-Varying Systems with Production and Environmental Risks

点击翻译按钮获取中文摘要

Reliability is the probability that a component or system will perform its intended function above a minimum standard for a specified period of time in defined circumstances. Research on modeling system reliability concerns reducing the effect of equipment and process faults on safety, process performance, and the environment. Good models of system reliability describe the effect of the process on damage accumulation (due to mechanisms that eventually make the system unreliable), using observable features of the system, to reduce uncertainty and operational risk. The goal for system reliability is to choose the lowest risk option based on the observed system outputs and conditional probabilities of what state the system is in, given the observed data. In this program, models will combine empirical operating & maintenance knowledge, failure physics, and data-driven models to yield more accurate machinery fault detection, identification, and prognostics in time-varying service, where faults can affect process performance, environmental impacts, and safety. Two inter-related investigations will be undertaken, with applications to reducing the impact of oilsands operations.