Helicopter gearbox fault detection and diagnosis using analog neural networks

This paper summarizes the results of two neural hardware implementations of a helicopter gearbox health monitoring system (HMS). Our first hybrid approach and implementation to fault diagnosis is outlined, and our results are summarized using three levels of fault characterization: fault detection (fault or no fault), classification (gear or bearing fault), and identification (fault sub-classes). Our second all-analog implementation exploits the ability, of analog neural hardware to compute the discrete Fourier transform (DFT) as a pre-processor to a neural classifier. Our hardware results compare well with previously published software simulations.<>