STRUCTURE-BASED CONNECTIONIST NETWORK FOR FAULT-DIAGNOSIS OF HELICOPTER GEAR BOXES

A new method of diagnosis is introduced for helicopter gearboxes that relies on the knowledge of the gearbox ''structure'' and characteristi cs of the ''features'' of vibration for component fault isolation. Bot h the structural knowledge and featural knowledge in this method are d efined as the fuzzy weights of a connectionist network that maps each sampled set of vibration features, obtained from a signal analyzer int o fault possibility values associated with individual gearbox componen ts. The structural weights in this network are defined to represent th e influence of gearbox component faults on the overall vibration sense d by each accelerometer, and the featural weights are defined to denot e the influence of components faults on individual vibration features. Given the extremely complex structure of helicopter gearboxes which p rohibits accurate modeling of the effect of faults on their vibration, the structural weights in this method are defined based on the root m ean square value of the frequency response of a simplified lumped-mass model of the gear-box. The experimental evaluation of the method base d on vibration data from two different gearboxes is included in a sepa rate paper.