Abstract: In order to improve the demodulation accuracy of cab signals, this paper proposed a cab signal demodulation method based on parameter optimized Variational Mode Decomposition (VMD) and Hilbert transform. The paper used the Least Mean Square (LMS) adaptive filter to filter the signals collected by the locomotive antenna, used correlation coefficient as the evaluation index, optimized VMD parameters, and used VMD to perform modal decomposition on the filtered signal, applied Hilbert transform to each Intrinsic Mode Function (IMF), constructed complex analytical expressions, and calculated the carrier frequency and low-frequency modulation signal frequency of cab signals, verified the feasibility and demodulation accuracy of this method in harmonic environments through computer simulation. Finally, the paper built a hardware platform for program execution based on the Cortex-M4 core microcontroller, build a cab signal generation circuit based on the Direct Digital Synthesis (DDS) device AD9854. This circuit generated orthogonal FSK signals in Ramped Frequency Shift Keying (FSK) mode to simulate the on-site cab signal testing environment. Through hardware testing, under low signal-to-noise ratio conditions, the maximum error of carrier frequency does not exceed 0.1 Hz, and the maximum error of modulation low-frequency demodulation does not exceed 0.05 Hz. The experimental results show that this method has strong resistance to harmonic interference and can accurately demodulate cab signals.