Detecting and isolating faults in complex engineering systems is important for properly planning maintenance, and it leads to decreases in down and repair times and to an increase in system performance. Mobile hydraulic valves, which are characterized by functions such as pressure-compensation and pilot-operation, are such complex systems that they could tremendously benefit from a real-time, accurate fault detection and isolation system. However, the complexity of these valves means that accurate full-state modeling is generally difficult and time-consuming. This paper proposes a reduced-order model for the fault detection of an open-loop-controlled mobile hydraulic valve. This reduced-order model is used along with statistically computed adaptive thresholds for the purpose of enhancing the reliability of fault detection. The considered faults include valve spool jamming caused by hydraulic fluid impurities, leakages caused by wear-induced increased clearances between the valve spool and sleeve, and sensor faults. The reduced-order model omits the modeling of pressure compensator dynamics by using a measurement of the pressure compensator, and pilot pressure dynamics by measuring the pilot pressures that drive the main spool. Experimental results from a commercial mobile hydraulic valve controlling a 2-DOF hydraulic crane show the practicality of the reduced-order modeling and adaptive threshold arrangements in this fault detection task. As a downside, a reduced set of faults can be isolated with the reduced-order model, but as a future consideration a bigger set of fault patterns with a full model are given and compared with the results obtained.

This content is only available via PDF.
You do not currently have access to this content.