Fault detection and isolation in fluid power systems using a parametric estimation method

Since the states or the parameters of a monitored process are closer to the process faults in terms of signal flow, observer-based state and system parameter estimation theories, are well developed in fault detection and isolation (FDI) for linear time invariant (LTI) systems. However, for systems, where explicit models are difficult to derive, the functional redundancy based FDI strategy, derived for LTI systems, cannot be directly implemented. In this paper, a parametric estimation method for FDI is employed for an electro-hydraulic system. Based on a set of criteria, an auto-regressive model with exogenous input (ARX) is first selected to approximate the dynamic behavior of the system. Next, the relationship between system´s supply pressure and the coefficients of the ARX model is studied through experiments. It is shown that direct threshold checking on estimated model coefficients can be employed as a FDI strategy to detect and isolate faults originating from incorrect supplying pump pressure.