Hydraulic Systems Degradation Detection Using Sparse Sensors

Abstract

A common issue in hydraulic systems is the degradation of sub-components. This is an important issue because hydraulics are the backbone of manufacturing, construction, and aerospace. When designing a hydraulic system, the sub-component's reliability is the primary requirement in component selection, because they determine the system's overall reliability. But, how the system actually fails is probabilistic, which varies based on operational conditions. To measure failure in a physical system during operation a model must be developed that measures the change in degradation from its initial healthy state. The accuracy of the prediction depends on the dynamics of the system and the system's operational input space. This thesis will present an approach for measuring degradation in a hydraulic system by using a dynamic model's prediction error to classify between sub-component fault states. The purpose of an inline fault detection system is to quickly recognize a failure and to provide information to the maintenance group on which sub-component has failed.