Abstract: In the context of the information age, applying deep learning technology to the field of automatic modulation recognition has become one of the key research topics in the communication industry. In response to the problems of complex model structure and large parameters in current deep learning based wireless communication signal modulation recognition technology, this paper proposes a modulation recognition technology based on convolutional neural network and gated recurrent unit technology. Through this technology, complex features and single channel features of in-phase orthogonal signals can be extracted, and the number of parameters in the system can be reduced, which has good application effects.