In this paper, a health monitoring method of electric scooter system with intermittent fault and working condition dependent degradation coefficients is proposed. Firstly, the bond graph model of electric scooter system is established to derive the analytical redundancy relations, and an integrated fault signature matrix is built, by which the intermittent fault detection and isolation can be implemented. Secondly, a sector search pigeon-inspired optimization algorithm is developed for intermittent fault estimation. After that, the fault duration feature is extracted with the aid of tumbling window, and a degradation model that includes working condition dependent degradation coefficients is proposed to describe the evolution trend of fault duration feature, based on which the remaining useful life of the faulty component can be predicted. Finally, the simulation results verify the effectiveness of the proposed method.