【作者】 孙书学；

【导师】 顾晓辉；

【作者基本信息】 南京理工大学 ， 火炮、自动武器与弹药工程， 2009， 博士

【摘要】 该文以智能反坦克子弹药(brainpower antitank submunition,简称BATsubmunition)为应用背景,具体研究运动声阵列对战场环境下地面声目标定向理论、跟踪理论及应用技术。全面论述了运动声阵列所涉及的基本问题,包括信号预处理、阵形研究、影响定位精度的因素等,重点研究了多尺度时延估计、运动声阵列定向算法以及BAT子弹药对地面机动目标角跟踪问题。主要研究内容简单归纳如下几点。首先,对声阵列定向涉及的基本理论方面进行了研究。概述声阵列的基本问题,结合全文的需求建立运动声阵列理论体系并分析了BAT应用中涉及到的一些基本问题,主要包括声源模型、声速简化计算模型、非直线传播修正模型、跟踪滞后修正模型、空气对不同频率声信号的吸收等问题;针对多传声器性能不一致问题进行研究,给出自适应识别性能较高的传声器的方法;对声阵列信号处理等问题也进行了研究,给出了一些有效的信号预处理方法及后处理方法,增强了声阵列定向稳定性,这是其它研究内容的基础和前提。其次,对声阵列定位基本原理及阵列定位能力进行研究,既能为BAT选择合适的阵型又可以为研究提高运动声阵列定向方法奠定基础。分析了声阵列定位的基本原理,指出声阵列定向与定距之间的关系,并推导了任意阵列定向与定距关系表达式,指出了提高定向精度即是提高定距精度的基本关系;在统一的条件下对比了各种阵型定向能力;提出正四棱锥形声阵列最小二乘定位算法,分析了正四棱锥形声阵列定向及定距误差;用试验对以上结论进行了验证;根据声阵列理论的研究结果以声学靶为背景讨论了多靶联合定位原理,从而在不用其它测速装置的前提下实现一层靶同时测量弹着点坐标和过靶速度。最后,根据分析,得出正四方形阵列较适BAT应用的结论。再次,对时延估计理论进行了研究。提出了基于经验模式分解(EMD)的多尺度时延估计技术,将信号分解为多个IMF进行互相关,形成多尺度时延值,并提出频域波形相似准则和时延矢量匹配准则,使得能够自适应的从多尺度时延值中选出精确的时延,使多尺度时延估计方法具有了实用性,另外改进和完善了基于小波分解的多尺度时延方法,从数学上证明了小波分解不会改变信号固有时延值的结论,使得小波分解的多尺度方法在时延估计中的应用具有了明确数学依据,并利用时延矢量匹配准则提取了精确时延值。试验证明:多尺度分解方法能保留信号中的高频成分,提高了时延估计精度,另外,该方法能够处理非平稳、且各通道间存在相关性噪声的信号,提高了该方法的适用范围。然后,重点对运动声阵列定向理论进行了研究。提出运动声阵列概念,通过理论分析将运动声阵列分解为平动声阵列和旋转声阵列,提出了“去平动”算法,给出了定向公式及其精度分析公式,并对平动声阵列定向进行了仿真研究,解决了声阵列平动过程中对目标定向问题;发现旋转声阵列是一种非线性系统,旋转速度对定向影响严重,为解决旋转条件下对目标定向问题,提出“去旋转”算法,给出了定向公式及其精度分析公式,并对旋转声阵列定向进行了仿真研究,结果表明,该算法放宽了受旋转速度的限制。最后,对BAT子弹药角跟踪问题进行了研究。根据仅有的声阵列估计出的带有误差的目标方向信息序列,推导了BAT对地面静止、运动目标距离估计公式,使得在未知弹体高度条件下根据方向信息即可估计目标全部的位置信息,目标距离的估计是进一步跟踪目标的前提条件,在此基础上研究了三维运动载体对二维机动目标的跟踪方法,提出了一种角度信息有误差的条件下角跟踪方法,并给出了可以对目标运动状态全部信息跟踪的充分条件。最后针对角信息序列相对角跟踪步长有冗余的矛盾将多尺度贯序式卡尔曼滤波方法引入角跟踪理论中来,有效解决了该矛盾提高了跟踪精度。更多还原

【Abstract】 Based on the application background of brainpower antitank submunition, the theories and applications of localization and tracking of acoustic target on the ground are studied with motion acoustic array. All basic problems involved by motion acoustic array are discussed, including preprocessing of signal, study on acoustic array shape, the elements which affects the localization accuracy, etc. Multi-scale time delay estimation(TDE), localization algorithm of motion acoustic array and bearings-only tracking of maneuvering targets on ground with BAT are studied on emphasis. The main research contents are as follows.The fundamental localization theories involved by acoustic array are studied at first. The basical problems of acoustic array are summarized. The theoretical system of motion acoustic array is established combining the requirement of the paper. Some elemental problems in BAT application are analyzed, such as source model, simplified algorithm model of acoustic velocity, nonlinear modify model of propagation velocity, delay modify model of tracking, absorption to different frequencies of air, etc. Aiming at un-consistent characteristics among different microphones, a self-adjusted recognition method for higher characteristic microphone is given. Signal processing of acoustic array is studied, some effective preprocessing and post processing methods of signal are proposed, which enhances the localization stability of acoustic array and becomes the basis and premise of following studies.The essential theory of localization and localization ability of acoustic array are studied, which provides BAT for selecting proper acoustic array shape and establishes the basis for improving the localization algorithm of motion acoustic array. The essential principle of localization of acoustic array is analyzed. Given the relationship of localization and fixed distance of acoustic array, the relationship formula of any array is deduced, which presents that to improve localization accuracy is to improve fixed distance accuracy. The localization abilities of different array shape are compared with the same conditions. The least square localization algorithm of pyramid acoustic array is proposed. The errors of localization and fixed distance of pyramid acoustic array are analyzed, which is verified by experiment. On the basis of acoustic array theory, the multi-target combining localization theory is discussed on the ground of acoustic target, which can measure the coordinate of spot and velocity. According to above analysis, squared array is proper for BAT.The TDE theory is studied. The multi-scale TDE based on empirical mode decomposition(EMD) is proposed, which decomposes signal as plenty of IMFs. The multi-scale time delay values are got by cross-correlation of different IMF. The similar criterion of wave in frequency(SCWF) and vector matching criterion of time delay estimation(VMCTDE) are proposed for selecting the accuracy TDE from plenty of multi-scale time delay values auto-adapted, which makes the multi-scale time delay applicable. The multi-scale time delay based on wavelet decomposition is improved and perfective. The conclusion that wavelet decomposition will not change the intrinsic time delay values of signal is proved, which gives specific mathematic principles for multi-scale method based on wavelet decomposition in application of TDE. The accuracy time delay value can be got by VMCTDE. The experiments present that multi-scale decomposition can remain the high-frequency of signal and enhance TDE accuracy, which can be applied broadly when the signal is nonlinear and the correlation noises exists among different channels.The localization theory of motion acoustic array is studied on emphasis. The concept of motion acoustic array is proposed. The motion acoustic array is decomposed into maneuvering acoustic array and rotating acoustic array. The de-maneuvering localization algorithm of maneuvering acoustic array is proposed. The localization formula and accuracy analysis formula are given. The simulation test presents that this algorithm can be applied in the localization of maneuvering target Because rotating acoustic array is nonlinear and rotating velocity affects localization accuracy badly, the de-rotating localization algorithm is proposed. The localization formula and accuracy analysis formula of de-rotating localization algorithm are given. The simulation presents that this algorithm reduces the limit of rotating velocity in some extend.The bearings-only tracking of BAT submunition is studied. The bearing information series with error is estimated with only acoustic array. The distance estimation formulas of BAT when targets are static or moving on the ground are deduced. Position information of target can be estimated based on bearing information series and unknown height of projectile. The estimation of distance is premise of tracking further. On the basis of above researches, the tracking method of 2-D maneuvering target is studied with 3-D motion carrier. The bearings-only tracking algorithm is proposed when the angle has error. The sufficient conditions of tracking all information of the target are given. Because of redundance confilict of angle information series to bearings-only tracking, multi-scale sequential Kalman filter is introduced to bearings-only tracking theory, which solves the conflict and enhances tracking accuracy.更多还原