Application of a PHM-based reliability prediction for an UAV’s control surface actuation system

Unmanned Aerial Vehicles (UAVs) become more and more important for the civil sector as they foster new business models and promise cost-savings for many applications. However, the acceptance and successful integration of this new aircraft type into the existing airspace largely depends on the overall system safety. From an operator’s point of view, it is therefore interesting to get a feedback on the current and future health status of flight critical systems. Current methods used to obtain and maintain safety constraints are based on reliability and safety engineering methods, which largely rely on the statistical processing of historic failure events to estimate the mean time to failure. In addition to this, modern systems are often equipped with Fault Detection and Isolation mechanisms. While the first approach is rather static and does not consider any feedback from an individual system (i.e. the actual health state), the latter is unable to give any predictions about the expected time to failure. With the discipline of Prognostics and Health Management (PHM) there exists an alternative approach to overcome these limitations. By assessing the actual degradation state of a system from sensor data (diagnosis) and predicting it (prognosis), the current and future health state can be estimated during flight. By further combining PHM results with conventional reliability methods, an improved prediction up to system level can be achieved. In this paper, the application of a PHM-based system reliability prediction is described. The system under study is the control surface actuation system for a hybrid UAV. By aggregating the Remaining Useful Lifetime (RUL) estimations of all monitored components, the future health states and system capabilities become predictable. By comparing these to the mission objectives the UAV’s operator gets a precise information on mission safety and is supported in the decision making process. The applicability of this approach is shown for a representative UAV mission scenario with exemplary RUL data. This paper is considered as a contribution to the PHM research at the end of the OSA-CBM process to connect RUL predictions with the decision-making process