运动目标激光微多普勒效应平动补偿和微动参数估计

利用激光探测微多普勒效应可以精确估计微动参数,有利于实现目标的准确分类和精细识别.运动目标的微多普勒效应是一种由多项式相位信号模型与正弦调频模型组成的混合信号.对于这类混合信号中的微动参数估计目前还未提出有效的方法.对此,本文提出一种基于分数阶傅里叶变换(Fr FT)的平动补偿方法,通过设计对Fr FT参数域的带宽搜索方法,可以从混合信号中精确估计平动参数,实现平动和微动的分离;通过设计静态参数粒子滤波器,从补偿后的信号中准确估计了微动参数;针对静态参数模型,采用马尔可夫-蒙特卡罗方法增加粒子多样性,并利用累积残差定义新的粒子权重计算函数,保证了算法在对多维参数估计时的快速有效收敛,避免了参数分别估计时误差传递的影响.通过仿真分析对比和实验数据,验证了本文所提补偿和参数估计算法的有效性.

Translation compensation and micro-motion parameter estimation of laser micro-Doppler effect

Precise target identification is significant for commanding and identifying enemies. The micro-Doppler effect (MDE) can reflect the subtle movement characteristics of the targets, which provides a new way of detecting and recognizing the target. However, the current research mainly focuses on the micro-motion feature extraction and classification of the targets, which is not capable of identifying the targets of the same type. In fact, by accurately estimating the micro-motion parameters and combining sufficient prior knowledge, the target can be accurately identified. Compared with the microwave radar, the laser detected MDE has high sensitivity and precision in micro-motion parameter estimation. This is more conducive to realizing the accurate classification and fine identification of the targets. In real detection, the MDE always exists in the moving targets. This will generate a mixed echo signal modeled by the polynomial phase signal and sinusoidal frequency modulation (SFM) signal. So far, there have been no effective methods of estimating the micro-motion parameters in such mixed signals. In this regard, a set of translational motion compensation and micro-motion parameter estimation methods is proposed in this paper. A bandwidth searching method based on the fractional Fourier transform (FrFT) is presented to precisely estimate the translation parameters, which will be used to realize the compensation for the translational motion. The advanced particle filtering (PF) method using the static parameter model is designed for the micro-motion parameters in the remaining SFM term. Given the lack of particle diversity in static parameter PF, the Markov chain Monte Carlo sampling is employed, which also helps to improve the algorithm efficiency. Meanwhile, a new likelihood function in calculating the particle weights is designed by using the cumulative residual. With this improvement, the correct convergence under multi-dimensional parameter condition is guaranteed. The proposed method can avoid the influence from error transfer and achieve efficient and accurate estimation. Compared with the typical method that combines the time-frequency analysis and the polynomial fitting through the simulation, the proposed FrFT method is verified to have little computation complexity and high estimation accuracy, where the relative estimation errors of the translational parameters are kept at 0.64% and 0.45%, respectively. The waveform similarity of the SFM signal phase between the compensated signal and the real one indicates that the accuracy fully meets the requirement for accurate estimation of the micro-motion parameters. Further, the simulation result also shows the high efficiency of the improved PF algorithm. The convergence time consumed by the proposed algorithm is 0.353 s, while the traditional method needs 0.844 s. In the end, the comparison with the experimental data from the traditional inverse Radon transform shows the effectiveness and necessity of the proposed method. The research results are conducive to the accurate and rapid estimation of micro-motion parameters, which lays a foundation for the fine target recognition based on the MDE.