FAULT DETECTION USING PARAMETER ESTIMATION FOR A HYDRAULIC SERVO-AXIS

A model-based, real-time capable fault detection and diagnosis system employing a Least-Squares parameter estimation approach is developed in this paper. A physical-based model of the proportional valve and the differential cylinder is derived. The performance of the current linearized flow model is evaluated and severe deficiencies are observed. Therefore, a nonlinear flow model is introduced, which governs the behavior of the hydraulic flow as a function of the valve spool displacement. Also, temperature effects are investigated and subsequently introduced into temperature dependent thresholds. Experimental results at a testbed with both intact and defective components show the good performance of the presented fault-detection and diagnosis method.