The use of pulsar X-ray signals for spacecraft navigation is a hot research field in deep space exploration. In X-ray Pulsarbased Navigation (XNAV), identifying X-ray pulsar signals is extremely important. This article proposes the Time Cycle Kurtosis (TCK) as an identification algorithm for extracting pulsar signal features, and combines it with ResNet34 for training and verifica‐ tion. The criterion for successful identification is based on an accuracy of over 90%, with the shortest observation times for B0531+ 21, B0540-69, and B1509-58 pulsars to complete the identification task being 0.5 s, 2.5 s, and 15 s, respectively. This article demon‐ strates through experiments that TCK can quickly and accurately complete the identification task of X-ray pulsar signals.