In order to realize the intelligent identification of signals modulation effectively, a novel modulation recognition method of MPSK(multiple phase shift keying）signals based on deep learning was proposed. The cycle spectrum of the MPSK signals were analyzed firstly, and the two-dimensional features information were obtained by extracting the contour map of the MPSK signals cyclic spectrum. Then, the two-dimensional features were trained by using the convolution neural network of deep learning. Finally, the effectiveness of the proposed modulation recognition method was verified by the test samples. The simulation results show that the proposed method has good recognition performance.