With the increasingly complex electromagnetic environment of communication, as well as the gradually increased radar signal types, how to effectively identify the types of radar signals is an important problem. To address this problem, a recognition framework based on deep learning and ensemble learning is proposed. This framework is composed of feature extraction and classifier design. In the first state, transform radar signals to time-frequency domain and learn the time-frequency picture feature by using stacked denoising autoencoders. A learning method of unsupervised pre-learning and supervise fine-tuning is used to train this deep model. In the second state, a model of ensemble multiple support vector machine classifier is created to recognize radar signals. Eight types of emitter signals are adopted in simulation experiment to validate the effectiveness of the proposed framework, and the results show that the joint model helps to obtain higher recognition accuracy.