【作者】 陈建春；

【导师】 杨万海；

【作者基本信息】 西安电子科技大学 ， 电路与系统， 2005， 博士

【摘要】 反辐射导弹(ARM)对防空雷达系统构成了严重威胁,在最近的几次局部战争中显示了其巨大的威力。有效对抗反辐射导弹的袭击,提高雷达自身的生存能力是决定现代战争胜负的一个关键因素。本文结合国防预研基金项目所开展的反辐射导弹检测算法研究,是对抗ARM的有效方式之一。针对现有脉冲体制跟踪雷达,研究了载机发射反辐射导弹初期,雷达自主进行反辐射导弹来袭告警的检测算法及实现问题。对反辐射导弹的检测可分为跟踪波门检测和截获波门检测两种情况处理。跟踪波门检测是利用雷达对载机进行正常跟踪状态下,目标距离跟踪波门内进行的检测,其特点是波门内的雷达回波信号为载机信号分量与ARM信号分量二者的合成,因此ARM的检测本质上为多分量信号检测问题。截获波门是专门为ARM检测而设置的距离门,其中的雷达回波信号为单一ARM信号分量。由于反辐射导弹在脱离载机的发射初期为朝向雷达站的匀加速运动,其回波信号为一线性调频信号,因此处于发射初期的反辐射导弹检测属于线性调频信号检测问题。本文对反辐射导弹检测研究的成果如下：建立了载机发射反辐射导弹的雷达回波解析信号模型。基于经典统计检测理论,建立了跟踪波门及截获波门内基于似然比／广义似然比检测的反辐射导弹最优检测统计量及最优检测系统的结构。截获波门检测的最优检测统计量为回波信号的加速度相位补偿(解线调)后的傅立叶变换。跟踪波门检测的最优检测统计量为抑制载机信号分量后的加速度相位补偿及傅立叶变换。研究了基于广义似然比最优检测统计量和基于时频分布的Radon-Wigner变换(RWT)及Radon-Ambiguity变换(RAT)的截获波门ARM检测算法。证明了基于RWT和直接RAT的检测统计量与基于似然比检测的最优检测统计量的同一性。根据瞬时自相关矩阵的结构特点,提出了一种瞬时自相关矩阵的构造方法,在不增加运算量的条件下使WVD和AF的输出信噪比提高了3dB,改善了检测性能。重点研究了距离跟踪波门内ARM的检测技术,提出了回波信号的载机信号分量抑制思想。针对时域CLEAN法抑制载机分量时存在的问题,提出了基于频域CLEAN法的频域白化CLEAN载机抑制方法。利用载机回波信号的谱特性设计载机信号白化滤波器,克服了时域CLEAN法对信号参数的敏感问题,提高了载机抑制算法的稳健性。根据维纳滤波理论,提出一种基于自适应线性预测误差滤波的载机抑制算法。利用回波信号的延时形式预测当前的载机信号分量,在线性最小均方意义下使得预测误差输出最小而达到载机信号分量的消除。该法充分利用了回波信号中载机信号自身的强相关性和ARM信号的弱相关性,使得载机信号分量被消除的同时ARM信号分量得以保留。该算法计算量小,易于实现。仿真表明该算法的载机信号抑制效果优良,在ARM信号分量低达-5dB的输入信噪比下,虚警及漏警的控制较为理想,原理上可在低信噪比情况下可靠地实现ARM的一发射即告警。利用小波包分解算法对载机信号分量抑制进行了尝试,载机抑制效果较为理想,但信号的小波包分解和重构计算量偏大。对两种波门下的检测统计量的检测性能从检测概率和输出信噪比两个方面进行了理论和定量分析。表明虽然在输出信噪比上基于线性包络的检测统计量与基于平方包络的检测统计量不同,但二者在奈曼-皮尔逊准则意义上其检测概率在给定的虚警概率条件下是完全相同的。针对目前防空雷达的现状,提出了现役雷达加装ARM检测告警设备的构想。针对基于自适应线性预测误差滤波结合加速度相位补偿频谱分析的ARM检测算法,给出了实现跟踪波门ARM检测和截获波门ARM检测的检测系统结构原理框图。对硬件及算法实现上的一些具体问题进行了分析,并对运算量影响较大的复量求模算法进行了研究,提出一种改进的近似求模算法,提高了模值计算的精度。更多还原

【Abstract】 Anti-radiation Missile (ARM) is a fatal threat to modern radar systems. It has shown great power in several recent regional conflicts. Developing counter-Anti-radiation missile techniques and preventing radar systems from being attacked by ARM is one of the key factors of leading to victory or defeat.The research work of this thesis is supported by the National Defence Foundation. Its main task is to develop practical algorithms for ARM launching detection. ARM launching warning is one of the effective measures for counter-Anti-radiation Missile. The optimum detectors with realization considerations are proposed in the thesis. The detectors are more suitable for pulsed tracking radars to detect an incoming Anti-radiation Missile at the first fire stage while the radar is in its tracking situation.ARM detection can be either made in the radar Range Gate (RG) or in an Interception Gate (IG) specially settled for ARM. Detection of ARM in RG is made by making use of the ordinary radar RG while the radar is in tracking of an aircraft. The radar echoes in RG are composed of two components. One is from aircraft, and the other is from ARM. Detection in this gate is essentially a multi-components signal detection problem. Whereas in IG, the radar echoes is only composed of ARM. In the first flying stage of an ARM, it moves toward the radar in a constant acceleration speed. The echoed signal from an ARM is a Linear Frequency Modulated (LFM) sinusoid wave. Hence, ARM detection is essentially LFM signal detection problems in lower signal to noise ratio.The main achievements of the thesis are as follow:The analytical radar signal model for ARM at its fire stage is established. The likelihood-ratio test based optimum detection criterion and decision-making rules for ARM detection in RG as well as in IG are constructed. The optimum statistical detector is configured based on the classic statistical detection theories. The optimum detection in IG is composed of Dechirping and Fourier transforming. However, in RG, an additional aircraft component canceller is included.The generalized likelihood ratio test based optimum detector, the Radon Wigner Transform (RWT) based detector and the Radon ambiguity transform (RAT) based detector for ARM detection in IG are analyzed. The identity in performance of the three detectors is demonstrated. Based on structure analysis of the simultaneous auto-correlation matrix, a modified method for the matrix construction is proposed. The result shows that a 3dB SNR gain can be obtained for WVD and AF, which improves the detection performance.ARM detection techniques in RG are discussed intensively and the idea of aircraft echo elimination is proposed. Because CLEAN in time domain is more sensitive to parameter estimation errors, a frequency domain CLEAN method called whiten filtering technique is introduced, which overcomes the shortage inherent in time domain CLEAN and improves the robustness of the detection algorithm.Based on the Wiener filtering theory, an adaptive aircraft echo canceller based on linear prediction error filtering (LPEF) is developed. The principle of the canceller is due to the auto-correlation of the input signals. With the method, the highly correlated aircraft echo with a mono frequency component can be eliminated perfectly while the feebly correlated ARM echo with a linearly modulated frequency can be hold. The advantages of this method are its less computation requirement and suitable for real-time processing. Computer simulation results show that reliable detection performance can be attained in lower signal to noise ratio. False alarm and miss detection is under better controlled. ARM echo signal can be detected immediately at the ARM’s first launch stage with this detector.Aircraft echo elimination based on Wavelet packet decomposition is proposed. Good elimination result obtained. However, the computation burden is great due to the wavelet packet decomposition and reconstruction of the signal.The performance of the proposed detector is analyzed quantitatively in detection ability and in output signal to noise ratio. The optimum linear envelope detector and the square envelope detector have essentially the same detection performance under the Neyman-Pearson Criterion in spite of the difference in their output SNR.Aimed at the present situation of our air defence radar systems, the idea of preventing radar systems against ARM attack by supplementing ARM detection units for current radar systems is introduced. The basic structure of the ARM detection unit based on the ALPEF with quadratic phase compensation Fourier analysis is given.Some problems related to the detection realization are considered. Complex modulus approximation, which is more computation wasted, is studied. The modifing CORDIC coefficients are given, which improves the precision of modulus approximation.更多还原