ABSTRACT
This chapter introduces a combined model-based/data-driven approach to failure prognosis that relies both on degradation models of the failing component and sequential Monte Carlo methods for state estimation. It provides an integrated failure prognosis architecture that is applicable to a variety of aircraft systems and industrial processes. The proposed particle filtering-based methodology addresses issues of uncertainty, system nonlinearity, and non-Gaussian noise. The prognosis task attempts to estimate how quickly the damage of an aircraft subsystem will progress. Modern aircraft/rotorcraft and critical industrial processes are equipped with monitoring, data acquisition, and data analysis hardware/software that are intended to assess the health of components/systems and inform the operator of impending failure conditions. The blind deconvolution denoising scheme was applied to vibration data derived from a faulted planetary gear plate, a critical helicopter component. The process of blind deconvolution attempts to restore the unknown vibration signal by estimating an inverse filter, which is related to partially known system characteristics.
