Unmanned aerial vehicles (UAV) collaborative edge computing can provide highly reliable and low-latency support for hot spot communication, emergency communication and other scenarios. Due to the load limitations of UAVs, an UAV equipped with a lightweight edge computing server is unable to handle a large number of computing tasks without cooperative offloading. This paper presents a joint optimization strategy based on the Deep Deterministic Policy Gradient (DDPG) algorithm, which minimizes system energy consumption by jointly optimizing offloading decision and computing resource allocation. Under the constraints of task delays and the number of access users, the joint optimization problem is transformed into a mixed-integer nonlinear optimization problem, and DDPG algorithm is utilized to train an appropriate offloading scheme and a computing resource allocation scheme to reduce system energy consumption. In the case of different number of users and different data sizes, the proposed strategy reduces the system energy consumption by 30%–45.8% compared with the random strategy.