In order to solve the problem of low efficiency and misrecognition of the mainstream method for individual signal recognition, a method for individual RF signal recognition based on residual reconstruction network is proposed. The frequency domain characteristics of the intercepted signal are obtained through Fourier transform and used as the input vector of the neural network; the residual network is reconstructed by using the advantage that the residual network can solve the problems of network degradation and gradient disappearance, and is used as a core network model for identifying individual radio frequency signals; and the number of model parameters is reduced by fixing the number of channels of each layer of network, so that the purpose of lightening the neural network is achieved. Experimental results show that compared with the ResNet18 method, the proposed method improves the individual recognition rate by about 3.8% for 30 target signals, and reduces the model size by 13 times, which can better solve the problem that the performance of model compression and recognition algorithm can not be balanced.