Research Of UAV Prognostic And Health Management Technology

With the development of UAV technology, modern equipment of UAV system has become increasingly complex, which presents new challenges to fault diagnosis and maintenance. Prognostic and health management technology can conduct a comprehensive health monitoring of UAV important component, and play an important role on safety performance improvement, support efficiency maintenace and life cycle cost reduction. Therefore, it has important theoretical meaning and application value for UAV development to research the UAV prognostic and health management (PHM) technology.In this paper, the main content is the core part of UAV PHM technology- Condition monitoring and fault prediction systems, in which fault prediction is the top priority. This paper applies data-based fault prediction algorithm on UAV system study. As the input and output response model is nonlinear and multi-parameters, we choose particle filter algorithm which not restricted to linear and Gaussian distribution, to approximate the implied fault mapping mechanism of the system data, and do fault forecast for providing decision-making of system maintenance and real-time command. UAV is an extremely complex system, in this paper, four aspects is focused on:1) Based on the data derived from a large high-altitude UAV, model the key parameters of UAV turboprop engine, then select particle filter for fault prediction.2) Based on the data derived from engine starting test, model the key parameters of turbojet engine starting process, then select particle filter fault prediction.3) Based on the data derived from a large high-altitude UAV, select attitude angle as research object for its significant effect to flight path, after modeling it, use particle filter to indirectly achieve the flight path prediction.4) Apply Condition monitoring and Fault prediction systems to the UAV support decision system of ground control station, verify its validity.Compared with the real UAV flying values, the prediction value got from fault prediction method based on particle filter can be a good fit to the real parameters. Therefore, it can be concluded that, in the actual UAV flight, after applying this system to the UAV support decision system, we can get the real-time prediction fault, timely remind the operator and provide complementary management recommendations. All these makes great help and significance to the UAV flight control.