As the Internet of Everything moves towards intelligent connectivity for everything, large amounts of data that need to be processed are being connected to the mobile internet. The integration of communication, computation and caching (3C) is the trend for next generation mobile communications. This integrated framework can dynamically coordinate 3C resources to meet the task demands of different multi devices (MD), effectively reducing the processing of duplicate content and the load on MEC data centers. However, such networks can consume a lot of energy. To meet the requirements of next-generation green wireless networks, this paper proposes an MD, multi-Unmanned Aerial Vehicle and multi-base station heterogeneous network 3C (MD-UAVs-BSsHN3C) integration framework to provide services for MD. Then, the energy consumption model of the 3C network is established. And then, the deep deterministic policy gradient (DDPG) algorithm is used to optimize the task execution location and network resource allocation of the MD-UAVs-BSsHN3C system to minimize system energy consumption under the constraint of delay. The performance is also compared with that of the Q-learning algorithm. Finally, the simulation results show that the system energy consumption of the model proposed in this paper is significantly better than that of the other three policy models. The impact of different MEC cache space variations on energy consumption reduction is further analysed.