In order to improve the traffic efficiency of freeway merging area under connected and autonomous environment, it is necessary to implement real-time cooperative control. Aiming at the traffic congestion problem of freeway merging area under medium and high flow densities, a grouping cooperative merging control method of connected and autonomous vehicle (CAV) is proposed taking into account of the access efficiency and model solving complexity. Firstly, designated detection zones of specific lengths are set up on the upstream section of the merging zone, and when the ungrouped vehicles are about to leave the detection area, all the ungrouped vehicles of the mainline and ramp in the detection area are grouped together based on the shortest time of the vehicles arriving at the merging point, which is based on K-means clustering. Then, the order of passage for different groups is determined based on the first in first out (FIFO) rule. Finally, a typical optimization model is sequentially applied to optimize the passage sequence and trajectory of vehicles within each group. A joint simulation platform using SUMO and Python is utilized to evaluate the effectiveness of cooperative control under different combinations of total flow rates of 1800, 2000, 2200 and 2400 vehicles per hour and flow ratios of 25:75, 50:50 and 75:25. The results show that compared with the grouping strategy based on headway threshold, the strategy in this paper can reduce the average delay by more than 4.5% and the computation time by more than 81.3% with the number of groups being 3. Group cooperative control of vehicles in the merging area can make a better balance between improving access efficiency and reducing computational complexity.