In order to accurately calculate the machine temperature distribution, maximize the machine performance and prevent the machine from overheating risk, a system-level multi-physical field prediction platform based on field-circuit coupling method is proposed in this paper. Firstly, the coupling effects among machine driver system, electromagnetic field, temperature field and fluid field are studied, and the coupling mechanism between multiple physical fields are revealed. Then, a system-level electromagnetic-temperature calculation model is established. Through several iterations, an accurate temperature distribution can be obtained. Motor temperature is monitored by drive system and armature current is adjusted in real time, which prevents the machine from overheating. Finally, the correctness of the model is verified by experiment.