Fault diagnosis of a complex mechanical structure helicopter main gearbox

Abstract

Reliable diagnostics capability is a bottleneck problem for complex mechanical structures due to a lack of proper online sensors. Embedded diagnostics (ED) has emerged as a critical enabler to provide real-time health state of the complex mechanical systems by integrating miniature online sensors into the structure. This thesis provides a detailed fault diagnosis approach of a complex mechanical structure– helicopter main gearbox using an embedded diagnostics system. The introduction of the multidisciplinary nature of the complex structure for fault diagnosis stands in need as the structure contains several coupled subsystems as well as sub-subsystems. Therefore, the sensor placement in gear, and bearing at the design stage is formulated as a multidisciplinary design optimization (MDO) problem. Among many solution methods, the deterministic multidisciplinary feasible (MDF) and robustness-based multidisciplinary feasible (RDO/MDF) approaches are used in this thesis to solve the multidisciplinary design optimization problem. The optimal solutions obtained from the MDO formulations give the best available values of the design variables that help the designer to place the sensors in the gearbox at the design stage. The vibration signals are achieved from the optimum location of the sensors via simulation as these signals are sensitive to the existence of the fault. The raw signals are post-processed using signal processing technique-filtering and supervised classification technique- support vector classification for fault detection. The achieved classification accuracy indicates that the proposed approach is highly reliable and applicable in fault diagnosis of the helicopter main gearbox system. The classified faults provide sufficient condition monitoring information of the system that may avoid any major breakdown of the structure.