Abstract: In view of the fact that active frequency response (abbr. AFR) of bulk power system is facing with a huge number of operation scenarios and its control strategy involves numerous factors, it is necessary to cluster its operation scenarios to improve the control efficiency on the promise of ensuring control accuracy. For this reason, a typical scene generation method of active frequency response control based on the improved fuzzy C-means clustering was proposed. Firstly, in allusion to such defects as complex solving process and difficulty of online solution, an estimation method of system frequency nadir under traditional frequency response control mode and active frequency response control mode was proposed. Secondly, according to clustering effectiveness index, the fuzzy C-means clustering algorithm was improved to calculate the number of scene clusters, and the membership function between the running scenes and the category was established. Finally, to ensure the security of system frequency, the worst operation scenario within the class was taken as a typical scenario, and an active frequency response control strategy for this scenario was enacted to conduct the off-line verification and on-line application. Results of simulation analysis show that the proposed method is feasible and effective in the stage of off-line analysis on active frequency response control and in the stage of online application.