For automotive manufacturers, intelligent fault diagnosis and prognosis techniques to ensure reliability of their products become increasingly important with the introduction of Industry 4.0. In this paper, an intelligent intermittent fault diagnosis and prognosis method is proposed for steer-by-wire (SBW) system based on composite degradation model. First, a nonlinear bond graph model of the SBW system is established, by which an integrated fault signature matrix combining analytical redundancy relations and dedicated observers is developed to enhance the fault isolability. Second, in order to identify the features (i.e., fault appearing and disappearing instants) of each intermittent fault, an improved whale optimization algorithm is proposed by introducing nonlinear convergence factor, Cauchy-Gaussian mixture mutation and greedy selection strategy. After that, the composite degradation model is established to predict the remaining useful life of the component with intermittent fault, where duration feature and frequency feature are considered with the aid of observation window. Finally, the effectiveness of the proposed approach is validated by simulation and experimental results.