Abstract: In this paper, a power optimization algorithm is proposed for the two-stage NOMA(Non-Orthogonal Multiple Access)-D2D(Device-to-Device) cooperative relay scene, and the outage probability of each user is deduced and analyzed. An optimization problem aiming at maximizing the ergodic capacity of the system is constructed, due to the difference of communication users in the first and second stages, the power distribution factor in the first stage which plays a decisive role in improving system capacity is optimized. The extreme point of the original optimization problem is proved, the optimal power distribution factor is obtained by using the property of convex function, and the outage probability of each user is further deduced and analyzed. The simulation results show that the theoretical outage probability value is completely consistent with the simulation results, which proves the correctness of the derivation. The power optimization algorithm proposed in this paper can significantly improve the system capacity and reduce the outage probability. For example, when SNR is 35dB, the algorithm in this paper can improve about 17.6% of the system capacity, while the outage probability of cooperative relay user is reduced by about 81.3%.