To monitor the realtime hydraulic pump run status more timely and accurately and to make fault diagnosis, a pump fault diagnosis vehicle terminal was built. In this terminal, an embedded processor S3C6410 was taken as the core hardware platform. The fault diagnosis system was constructed based on BP neural network with embedded Linux as operating system and Qt4 as application framework. The hydraulic pump operating conditions realtime monitoring, realtime fault warning and fault diagnosis functions were realized.It was proved that the parameters of BP neural network could be updated by 3G network, so as to improve the fault diagnosis ability of the system.