The traditional steering engine fault detection was usually based on the known parameters of steering engine, but in practice, most of the internal parameters of the steering engine were unknown. A Kalman filter for filtering fault detection method based on subspace identification was proposed with the thought of establishing fault observer in this paper. At the same time, based on hardware-in-the-loop simulation tool dSPACE, a semi-phisical simulation platform for electric steering engine fault detection was set up, through which the fault on-line detection and fast fault analysis of unknown-parameters steering engine were realized. The experimental results showed that this method could not only detect the fault occurence time effectively, but also identify the fault types and degree quickly, and the achievement provided a solution for the engineering application rapid fault detection of steering engine in UAV.