Abstract: Based on training neuroidentifier in off-line way in active control of air suspension system, BP artificial neural network was applied in the research of identifying dynamic load of air suspension system by vibrant acceleration of under spring mass. A dynamic model of 1/4 engineering vehicle with air suspension system was built. Vibrant acceleration data of under spring mass and dynamic load data of engineering vehicle were acquired by simulation. Vibrant acceleration data of under spring mass were input data of BP neural network, dynamic load data were output data of BP neural network. BP neural network was trained by input and output data of air suspension system. Generalized ability of trained BP neural network was tested. Road input was sine wave that its amplitude was 0.01 m and its frequency was 1 rad/s. The percentage was 82.95% that the ratio of failure identifying points to total points less than 30%. Road input was sine wave that its amplitude was 0.02 m and its frequency was 2 rad/s. The percentage was 77.94% that the ratio of failure identifying points to total points less than 30%. The result indicates that BP neural network can adapt different road inputs.