A MODULAR INTELLIGENT DATA ADMINISTRATION APPROACH FOR HELICOPTER HEALTH AND USAGE MONITORING SYSTEMS

Much effort and expenditure has been focused on the introduction of ac quisition and processing hardware for helicopter health and usage moni toring systems (HUMS) operations. In contrast, the potential difficult ies in the management of HUMS data have received much less attention. HUMS hardware will typically generate in excess of one megabyte of dat a per Fight. This data will be downloaded to ground station computers for further analysis. A modular intelligent data administration system (MIDAS) is highly desirable in order to efficiently extract useful in formation from the huge amount of HUMS data, and provide general robus t diagnostic strategies. A state-of-the-art implementation of relevant technologies and analysis methods is essential to realize such a syst em. This paper presents the requirements and structure of MIDAS and de tails an approach that integrates neural network technology, helicopte r mathematical dynamic models and unsupervised pattern recognition tec hniques. Examples using vibration and spectrometric oil analysis progr ammes (SOAP) as a means to evaluate the health and usage of helicopter components are presented.