基于粒子滤波的汽车防撞雷达研究

【作者】 杜炳硕；

【导师】 宋耀良；

【作者基本信息】 南京理工大学 ， 通信与信息系统， 2004， 硕士

【摘要】 毫米波汽车防撞雷达是目前国内外正在研发的汽车安全系统的重要组成部分。由于工作于各种非高斯背景的干扰环境，利用以往传统的Kalman和扩展Kalman滤波等线性或近似线性的处理方法很难实现对目标的准确检测和跟踪。粒子滤波技术对于非线性系统和非高斯的干扰环境具有高度的适应性，可以很好解决以往算法精度不高甚至发散的问题，克服系统可靠性差的缺点。研究基于非线性和非高斯背景下的粒子滤波信号处理技术对于提高毫米波汽车防撞雷达性能，降低恶性交通事故的发生率，提高交通安全性有着极其重要的理论和现实意义。本文主要围绕基于粒子滤波技术的汽车防撞雷达开展工作，主要内容包括： 1．在介绍非线性滤波的基础上，引入粒子滤波，并通过实例对粒子滤波性能进行仿真，结果证明了在非线性系统非高斯背景下粒子滤波的优异性能； 2．简要介绍了前视车载雷达的预警功能、特性指标，分析了车载毫米波雷达系统的工作体制和天线的扫描方式，提出了采用混沌调频信号代替传统的线性调频信号的方案，通过仿真表明该信号在很大程度上可以减少由各种背景干扰及交叉干扰引起的虚警现象； 3．构造了整个车载雷达系统的原理框图，对其工作原理进行了讨论。通过分析和仿真证明了对单目标和多目标相关处理具有良好的测距性能；简单介绍了数字多波束技术和测角技术，仿真结果显示这种高分辨阵列测向技术具有高精度； 4．分析了粒子滤波在汽车防撞雷达系统中的应用，并与扩展卡尔曼滤波进行比较，仿真结果表明：粒子滤波很好地解决了扩展卡尔曼滤波跟踪精度不高甚至发散的问题，大大地提高了系统可靠性。更多还原

【Abstract】 The millimeter-wave band automotive collision warning radar, which has been developing all over the world, is an important part of vehicle safety system. Because the radar is applied in all kinds of non-Gaussian interference circumstances, it is difficult to detect and track the target by using traditional Kalman and Extended Kalman methods which were liner or approximated linear measures. Particle filter is very fit for non-linear and non-Gaussian interference circumstances. It has a high-veracity. Such a vehicle radar used particle filter in practical transportation application can provide great benefits in safety and driver convenience. It has academic and practical significance. Research works and the major contributions in this thesis are as follows:1. After discussing non-linear filter, particle filter is introduced. Both the theoretic analysis and simulation results show that the particle filter does well at a non-linear and non-Gaussian interference circumstances.2. After analyzing the functions and specifications of the system, discussing radar working method and antenna scanning mode, the scheme using chaotic FM signal as a substitute of LFMCW signal is proposed, the results of analysis and simulation show that chaotic FM signal is superior to the traditional LFMCW signal in electromagnetic compatibility. It can solve the false alarm problem to some extent.3. The principle block diagram of the vehicle radar system is given. The simulation results of correlation show that it has a good resolution of distance. The azimuth angle measurement problem of automotive collision warning radar is discussed. DBF is introduced. The simulation shows it has a good split-angle precision.4. The application of particle filter in the millimeter-wave band automotive collision warning radar is analyzed, the simulation show that the performance of particle filter is superior to extend Kalman, it can solve the problem of low- precision and low-reliability.更多还原