Abstract：In the direct-drive multi-car elevator woking process, passenger traffic mode is an important factor affecting the performance of the elevator system. In order to improve the service standards of the system, a reasonable strategy of group control scheduling was formulated to realize the optimal scheduling of the system. A direct-drive multi-car elevator group control system based on destination reservation was designed. In the case of predicting passenger traffic information in advance, the fuzzy logic control was used to identify the passenger traffic of the multi-car elevator system by establishing membership function and fuzzy rules. Finally, the fuzzy control was compared with the BP neural network algorithm by application examples. The analysis showed that, in the direct-drive multi-car elevator group control system based on destination reservation, the fuzzy logical control was much more convenient and accurate to identify the different types of passenger traffics under the limited experimental samples, which layed a foundation for further optimal scheduling of the system.