With the development of aerospace technology，space has become a strategic focus，and the perception of spatial targets is particularly crucial，including the detection of satellite electromagnetic spectra and individual satellite identification.Telemetry signals are common radiation source signals for satellites in orbit，and individual satellite targets can be effectively identified by in-depth analyses on the potential features in their telemetry signals.In this paper，an identification method based on multi-modal fusion is proposed.It combines residual convolutional networks and long short-term memory （LSTM） networks，and can enhance the network's ability to extract and fuse deep features of signal data，significantly improving the identification accuracy.Experimental tests show that the proposed method achieves an average identification accuracy of 92% on the actual data set.This study has guiding significance for the development of satellite individual identification technology，and is expected to promote further research and application in this field.