To address the under?sampling issue in blade tip?timing signals and overcome probe limitations, a novel method for extracting blade natural frequencies using single?sensor measurement data is proposed. By constructing multiple virtual signals to frequency?shift the measured signal, the frequency components of dynamically varying signals are extracted, and the natural frequency information of the blades is retrieved in conjunction with prior knowledge. Simulation and experimental results validated the effectiveness and accuracy of this method in natural frequency extraction, and also demonstrated the potential for processing small?sample data. This approach uses only a single blade tip?timing probe, overcoming the restrictions in probe placement. With the minimum number of probes, it resolves the frequency aliasing caused by under?sampling, offering the promising technical support for online monitoring of rotor blades in aero?engines.