A Parameter Estimation Method for Precession Targets Through Improved GD-TLS

To address the limitations of existing algorithms of precession target parameter estimation, including inaccurate precession parameter estimation and incomplete modeling of impulsive noise, a precession target parameter estimation method employing synchronized wavelet transform (SWT) and an improved gradient-descent total least-squares (GD-TLS) algorithm is proposed in the paper. First, for the scenario where the radar line-of-sight angle is approximately 90°, the micro-Doppler signal characteristics of precession targets under both translational and non-translational motion are analyzed, and SWT is adopted to pre-estimate the precession frequency, avoiding the influence of time-frequency curve overlapping in short-time Fourier transform. Next, the improved singular value ratio spectrum method is utilized to accurately estimate the precession frequency through traversal search. Finally, the GD-TLS algorithm is optimized based on micro-Doppler derivation: an exponential function is introduced into the cost function to suppress impulsive noise, which enhances the algorithm′s immunity to impulsive noise during precession parameter estimation. Experimental results show that the proposed algorithm achieves an estimation error of 0.4 % under noise-free conditions, and still maintain an estimation error within 5.2 % under low signal-to-noise ratio (10 dB) with impulsive noise, outperforming traditional methods including the least mean square, recursive least squares, and robust least mean logarithmic square, which verifies the strong impulsive noise robustness of the proposed method.