Ubiquitous applications, especially in harsh environments and with strict safety requirements, make Fault Detection and Diagnosis (FDD) in hydraulic actuators an imperative concern for the industry. Model-based FDD uses estimation strategies, including observers and filters as estimation tools. In these methods, observability is a limiting factor in information extraction and parameter estimation for most applications such as in fluid power systems. To address the observability problem, adaptive strategies like Interactive Multiple Model (IMM) estimation have proven to be effective. In this paper a computationally efficient form of IMM referred to as the Updated IMM (UIMM) is used and applied to an Electro-Hydrostatic Actuator (EHA) for FDD. The UIMM is suited to fault conditions that are irreversible, meaning that if a fault happens it will persist in the system. In essence the UIMM follows through a progression of models that in line with the progression of the fault condition in lieu of having all models being considered at the same time (as is the case for IMM). Hence, UIMM significantly reduces the number of models running in parallel and at the same time. This has two major advantages which are higher computational efficiency and avoiding combinatorial explosion. The state and parameters estimation strategies that is used in conjunction with UIMM is the Variable Boundary Layer Smooth Variable Structure Filter (VBL-SVSF). The VBL SVSF is a robust optimal estimation strategy that is more stable than the Kalman Filter in relation to system and modeling uncertainties. The UIMM method is validated by simulation of fault conditions on an EHA.