Abstract: When processing micro-Doppler signals with relatively high micro-motion frequencies, the performance of existing modal decomposition algorithms is greatly reduced due to the limitations of interleaved signal properties and the empirical parameters of the algorithms. By utilizing the physical property that micro-Doppler signals can be expressed as multiple sinusoidal frequency modulation modes, combing nonlinear frequency modulation decomposition algorithms and high-precision time-frequency spectrum technology, a sinusoidal frequency modulation mode decomposition algorithm for micro-Doppler signals is proposed. Based on the sparsity of the micro-motion frequency of micro-Doppler signals, the algorithm constructs a fractional trigonometric function basis that includes micro-motion frequency estimation, thereby improving the estimation accuracy of the instantaneous frequency characteristics of multi-component micro-Doppler signals. Simulation results verify that the proposed sinusoidal frequency modulation mode decomposition algorithm achieves a 3 dB improvement in output signal-to-noise ratio (SNR) compared to traditional intrinsic chirp component decomposition algorithms and exhibits stronger universality for targets with different micro-motion frequencies.