Engine electric accessories are a promising technical approach to improve engine fuel efficiency. The engine’s electric cooling fan, however, with few signals to controllers, has a challenging fault detection issue. In this paper, a fault diagnosis algorithm based on model and support vector machine was proposed. Firstly, a dynamic model of electric motor-cooling fan system was established, then a real time model-based observer was designed to estimate the torque of the cooling fan. The load torque estimated with control signals was used to identify cooling fan model parameters. According to the identified parameters, a support vector machine (SVM) was utilized for fault classification . The simulation and experiment showed ,this diagnosis algorithm was able to discover over 98% mechanical failures with an classification accuracy of 95%.
Keywords Electric fan, fault diagnosis, support vector machine