Abstract: In view of the neglect of the interaction between push service and users during the process of “active perception-automatic generation-active push” in driving active service system (DASS), a driving push service platform (DPSP), which covers storage layer, application layer and evaluation layer, is designed based on the operation behavior of driving users. The storage layer is used to classify and store the pervious and current data of user behavior. In application layer, B-Num/BT algorithm is designed based on user's operation behavior, for improving the user acceptance of DPSP, while the evaluation layer utilizes operation data to evaluate and monitor the performance of push service. A driver-in-the-loop test is conducted to verify the timeliness, safety and accuracy of typical push service scenarios. The results show that DPSP meets the performance requirements of push system. In addition, it is found that the timeliness also has the function of classifying driving users. The research has great significance for the perfection of DASS system, the strengthening of driving safety and even the popularization of autonomous driving technology

Key words: driving active service, operation behavior prediction, driving push service platform, typical service scenarios