CN101644758B - An object localization and tracking system and method - Google Patents

An object localization and tracking system and method Download PDF

Info

Publication number
CN101644758B
CN101644758B CN 200910078474 CN200910078474A CN101644758B CN 101644758 B CN101644758 B CN 101644758B CN 200910078474 CN200910078474 CN 200910078474 CN 200910078474 A CN200910078474 A CN 200910078474A CN 101644758 B CN101644758 B CN 101644758B
Authority
CN
China
Prior art keywords
target
module
particle
state
cluster
Prior art date
Application number
CN 200910078474
Other languages
Chinese (zh)
Other versions
CN101644758A (en
Inventor
李宇
王彪
黄海宁
Original Assignee
中国科学院声学研究所
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中国科学院声学研究所 filed Critical 中国科学院声学研究所
Priority to CN 200910078474 priority Critical patent/CN101644758B/en
Publication of CN101644758A publication Critical patent/CN101644758A/en
Application granted granted Critical
Publication of CN101644758B publication Critical patent/CN101644758B/en

Links

Abstract

本发明涉及一种目标定位跟踪系统及方法。 The present invention relates to a method for target location and tracking system. 所述系统包括:多个簇定位模块及指控模块;所述簇定位模块包括:多个传感器节点及簇头节点;所述簇头节点包括:初始化模块、粒子滤波模块、粒子权值计算模块、重采样判断模块及估计目标状态模块。 Said system comprising: a positioning module and a plurality of clusters alleged module; locating the cluster module comprising: a plurality of sensor nodes and the head node of the cluster; the cluster head node comprising: an initialization module, particle filtering module, the particle weight value calculation module, resampling the target state determination module and estimation module. 本发明采用一种结合卡尔曼和粒子滤波的方法实现目标的被动定位,其在实现高精度定位的基础上,运算速度远远低于粒子滤波方法;采用多传感器基于对目标方位的观测最终实现对目标的定位跟踪,可以克服传统机载或舰载单站系统在观测期间必须进行机动的约束,不需要观测平台的机动,提高了目标定位的灵活性,大大增加目标监测定位的区域面积,避免存在定位盲区的不足;具有很高的有效性、精确性及可行性。 The present invention employs a method and Kalman particle filter Passive targeting, which is based on high precision positioning on the operating speed far below the particle filter; multi-sensor based on the observation of the target azimuth ultimately positioning of the target tracking, can overcome the traditional airborne or marine single station system must be constrained during the observation of a motor, the motor does not require the observation platform, improving the flexibility of positioning the target, greatly increasing the target area of ​​the monitoring location, avoid positioning the shortcomings of the blind; with high effectiveness, accuracy and feasibility.

Description

一种目标定位跟踪系统及方法 An object localization and tracking system and method

技术领域 FIELD

[0001] 本发明涉及信息技术领域,特别涉及一种目标定位跟踪系统及方法。 [0001] The present invention relates to the field of information technology, particularly to a method for targeting and tracking system. 背景技术 Background technique

[0002] 由于现代信息技术、网络技术及无线通信技术的发展,推动了无线传感器网络技术的迅速发展。 [0002] Since the development of modern information technology, network technology and wireless communication technology, to promote the rapid development of wireless sensor network technology. 运动目标的定位与跟踪技术具有广阔的应用背景,涉及到军事和民用领域。 Moving target location and tracking technology has broad application background, involving military and civilian fields. 传统的目标被动定位技术中,多采用基于纯方位测量的目标被动定位技术,其主要解决的问题是如何利用观测的目标方位信息来估计目标的运动参数进而实现目标的定位跟踪。 The traditional goal of passive positioning technology, the use of passive positioning technology based on pure target azimuth measurement, the main problem is how to use the observation target location information to estimate the motion parameters so as to realize the target location and tracking targets. 系统配置多为机载或舰载单站测量系统,该系统配置由于受到运动目标可观测性原理的限制,要实现对目标的定位,观测平台需要在观测期内进行机动,可有时候由于观测平台无法机动,近而不能实现目标的定位;无线传感器网络通过部署的多个传感器节点组成,通过无线通信方式形成的一个多跳的自组织的网络系统。 Multi-system is configured to airborne or marine single station measurement system, the system is configured to limit due to the moving object can be observed in principle, to achieve the positioning of the target, the platform needs to maneuver the observation in the observation period may be sometimes due to the observed not motorized platform, positioned near the target can not be achieved; wireless sensor networks via a plurality of sensor nodes deployed composition, a self-organizing network system formed by multi-hop wireless communication system. 其目的是协作地感知、采集和处理网络覆盖区域中感知对象的信息。 The aim is cooperatively sensing, collecting and processing information network coverage area of ​​the sensing target.

[0003] 从技术方面讲,基于纯方位目标运动分析问题本质上是一个非线性Bayes滤波问题。 [0003] From a technical perspective, based on the nature of bearings - only target motion analysis problem is a non-linear Bayes filtering problem. 由于非线性的原因,其精确解通常并不解析,所以工程上多采用基于扩展卡尔曼滤波(EKF)的次优解,它是利用泰勒级数展开,取其一阶近似使模型线性化,再利用卡尔曼滤波(KF)算法进行求解。 Due to the nonlinearity of the reasons, which are usually not accurate analytical solution, so the use of multiple sub-optimal solution based on Extended Kalman Filter (the EKF) in engineering, it is to use the Taylor series expansion, whichever is the first order approximation that the linear model, then Kalman filter (KF) algorithm to solve. 而且EKF滤波在模型非线性较强以及系统噪声非高斯时估计的精度严重降低,并可能造成滤波器的发散。 And filtering severely reduced EKF models strongly nonlinear and non-Gaussian noise estimation accuracy of the system, and may cause filter divergence. 而无味卡尔曼(UKF)采用UT(UnSCented Transformation)变化,虽然较EKF有更好的滤波效果,但都是在基于模型线性化和Gauss 假设的条件下。 And the unscented Kalman (UKF) using UT (UnSCented Transformation) changes, but are based on the linear model and the conditions assumed in Gauss EKF although better results than filtering. 因此人们一直在寻找适用于非线性非高斯Bayes滤波方法来提高估计精度。 So people are always looking for suitable nonlinear and non-Gaussian Bayes filtering method to improve the estimation accuracy. 近年来国际上在该方向的研究取得了令人瞩目的成就,其中尤以粒子滤波(PF)受到更为广泛的关注,所以近年来又出现了基于粒子滤波方法实现目标的纯方位测量定位技术, 该方法虽然比基于KF的方法定位精度高,但是该方法对实现后验概率的精确估计需要较多的粒子数,导致计算量大,计算时间长,不利于实时应用。 In recent years, international research in this direction has made remarkable achievements, especially in particle filter (PF) subject to wider attention, so in recent years has emerged pure azimuth measurement positioning technology to achieve the goal of particle filter based on this method, although high positioning accuracy than the method based on KF, but the methods to achieve accurate estimate of the posterior probability of the number of particles need more, leading to large computation, computing for a long time, is not conducive to real-time applications.

发明内容 SUMMARY

[0004] 为了克服对单站观测系统的不可观测性和粒子滤波计算速度慢这两个不足,本发明提供了一种目标定位跟踪系统及方法,该系统及方法基于上述对传统定位技术的分析基础之上,采用一种新的目标定位技术实现运动目标的分析,可以有效克服单站测量必须进行机动的不足,而且采用一种改进的粒子滤波方法实现定位,其计算时间远远低于采用粒子滤波实现定位的方法。 [0004] To overcome the non-observability and particle filter calculation speed slower single station observing systems lack both, the present invention provides a method for target location and tracking system, the system and method of analysis based on the conventional positioning techniques basis, using a new analysis technique targeting moving objects, a single station can overcome the lack of measurements must be carried out maneuver, and the use of an improved particle filtering to achieve positioning, using the calculation time is much less than particle filtering method of targeting.

[0005] 为了达到上述目的,本发明提供的一种目标定位跟踪系统,包括:多个簇定位模块及一指控模块。 [0005] To achieve the above object, the present invention provides a target position tracking system, comprising: a positioning module and a plurality of clusters charges module.

[0006] 所述簇定位模块包括:多个传感器节点及一簇头节点。 [0006] The cluster positioning module comprising: a plurality of sensor nodes and a cluster head node.

[0007] 所述传感器节点,用于对其所在簇内区域进行监测,当有目标出现,则对目标的方位进行测量,并将该测量值与自身坐标信息发送到本簇内的簇头节点。 [0007] The sensor node, for monitoring their location within the cluster region, when the target appears, then the azimuth of the target is measured, and the measured value is transmitted to the coordinate information itself within the cluster head node of the present cluster . [0008] 所述簇头节点,用于根据本簇内的传感器节点发送的测量值及其坐标信息,完成目标状态估计,实现目标的定位跟踪,并将定位跟踪结果发送到指控模块。 [0008] The cluster head, according to the measured values ​​and the coordinate information transmitted from the sensor nodes within this cluster, the target state estimation is completed, to achieve the target location tracking and location tracking module sends the result to the charges.

[0009] 所述指控模块,用于收集并显示各簇头节点的定位跟踪结果。 [0009] The charges module, for collecting and displaying the location tracking result of each cluster head.

[0010] 所述簇头节点进一步包括: [0010] The cluster head further comprising:

[0011] 一初始化模块,用于建立目标运动及观测方程、目标的初始状态分布函数和初始状态的估计方差,并随机产生N个粒子。 [0011] an initialization module, and for the establishment of the target observation equation, the variance of the initial estimate of the target state and initial state distribution function, and generate N random particles.

[0012] 一粒子滤波模块,用于利用状态方程对每个粒子状态进行基于扩展卡尔曼滤波的预测、更新。 [0012] a particle filter module, using the state equation for each particle based on the predicted state extended Kalman filter is updated.

[0013] 一粒子权值计算模块,用于计算粒子权值。 [0013] a particle weight calculation module for calculating the particle weight.

[0014] 一重采样判断模块,用于根据有效粒子数Nrff判断是否要重采样,若有效粒子数Neff小于预先设定值,则进行重采样,根据重要性密度函数重新采样N个粒子并分配权值。 [0014] Analyzing a resampling module for determining whether the number of effective particles Nrff be resampled, if the effective number Neff of particles less than the preset value, the re-sampling and re-sampling the N particles in accordance with the importance assigned a weight density function value. and

[0015] 一估计目标状态模块,用于根据粒子的权值计算估计目标状态。 [0015] a target state estimation module, for estimating the state of the target is calculated according to the weight of the particles.

[0016] 其中,所述初始化模块进一步包括:一建立目标运动及观测模型模块、一建立目标初始化状态及估计方差模块及一随机粒子产生模块。 [0016] wherein, the initialization module further comprises: establishing a target motion and the observation model module, and a goal of establishing an initialization state estimation module, and a random variance of particle generation module.

[0017] 所述建立目标运动及观测模型模块,用于建立目标的一阶马尔可夫系统方程及目标观测的观测模型: [0017] The establishment of the target model and the observation means for observing a target model order Markov system equation and the observation target:

[0018] [0018]

Figure CN101644758BD00061

[0019] 其中:XT = [xT,vx,yT,Vy]表示目标的运动状态,xT,yT分别表示目标在X轴及Y轴的坐标值,Vx, Vy分别表示目标在在X轴及Y轴的绝对速度分量;S'_=[<,^]i = 1,2…N, 表示传感器节点i分别在X轴及Y轴的坐标,N表示观测到目标的传感器节点个数;θ i表示目标与传感器节点的相对方位叫表示第i个传感器节点的测量噪声。 [0019] wherein: XT = [xT, vx, yT, Vy] represents the target state, xT, yT represent a target coordinate value X-axis and Y-axis, Vx, Vy denote target in the X-axis and Y the absolute velocity component of the shaft; S '_ = [<, ^] i = 1,2 ... N, represents a sensor node i are coordinates of the X-axis and Y-axis, N is the number of sensor nodes to the observed object; i [theta] It indicates the relative bearing sensor target node name indicates the i-th measurement noise sensor nodes.

[0020] 所述建立目标初始化状态及估计方差模块,用于建立目标的初始状态概率分布函数及估计方差。 [0020] The goal of establishing the initialization state and variance estimation module, for establishing the initial state probability distribution function of the target and the estimated variance.

[0021] 所述初始状态概率分布函数为:Xk+1 = FkXk+GkAk。 The [0021] initial state probability distribution function is: Xk + 1 = FkXk + GkAk.

[0022] 其中:Ak= [ax,k, 系统的处理噪声,即由于目标运动环境的不确定性导致分别在X和Y轴引起的加速度噪声,Ak〜N(0,Rk); [0022] wherein: processing the noise Ak = [ax, k, of the system, i.e., the target due to the uncertainty caused by movement in the environment causes each of the X and Y axes acceleration noise, Ak~N (0, Rk);

[0023] [0023]

Figure CN101644758BD00062

[0024] 所述随机粒子产生模块,用于从初始状态分布ρ (X0)中随机抽取N个初始粒子{χΛ i = 1,2···Ν},与初始粒子对应的协方差矩阵为:{ΡΛ i = 1,2…N}。 [0024] The random particle generating means for distribution ρ (X0) from the initial state of primary particles randomly selected from the N {χΛ i = 1,2 ··· Ν}, corresponding to the covariance matrix of primary particles: {ΡΛ i = 1,2 ... N}.

[0025] 其中,所述粒子滤波模块进一步包括:一预测模块和一更新模块。 [0025] wherein the particle filtering module further comprises: a prediction module and an updating module.

[0026] 所述预测模块,用于利用状态方程预测k时刻的目标的状态劣/4_,及预测协方差矩 [0026] The prediction module, a target state prediction using the equation of state in time k inferior / 4_, Covariance prediction and moments

阵4 其中=^L-, =FLX“+G'A +QLr FL =Fk Array 4 wherein = ^ L-, = FLX "+ G'A + QLr FL = Fk

[0027] 所述更新模块:用于利用k时刻的测量值对粒子预测结果及估计协方差矩阵进行更新,得到k时刻更新值及协方差矩阵Xki和ΡΛ [0027] The updating module: used to predict results and estimated particle covariance matrix updated using the measurement at time k, to obtain the updated value at time k and a covariance matrix and ΡΛ Xki

[0028]其中:xi = X1klk^ + KKyk -Kxiklk^)) Pl =(1-KikH•此k_〜· [0028] where: xi = X1klk ^ + KKyk -Kxiklk ^)) Pl = (1-KikH • this k_~ ·

J 9 J 9

jji dh. jji dh.

[0_ Hk = - U4t _ ^ = Plik^^^ + 。 [0_ Hk = - U4t _ ^ = Plik ^^^ +.

[0030] 其中,所述粒子权值计算模块根据贝叶斯及粒子滤波理论计算粒子权值: [0030] wherein the calculation module calculates the particle weight particle weight and the particle filter Bayesian Theory:

Γ0031ι a,' =iy' P(ZlJxik)MO . L0031」 Oh-CO1^ ' Γ0031ι a, '= iy' P (ZlJxik) MO. L0031 "Oh-CO1 ^ '

HK-^k ' -xOJcArOJc / HK- ^ k '-xOJcArOJc /

[0032] 其中:(1(¾/¾+/, z1:k)是重要性密度函数,p(zk/xkJ)是观测似然函数,P(XkAk^1) 是系统一阶马尔可夫过程的转移概率。 [0032] wherein: (1 (¾ / ¾ + /, z1: ​​k) the importance of the density function, p (zk / xkJ) an observation likelihood function, P (XkAk ^ 1) is a first-order Markov process system the transition probability.

[0033] 其中,所述有效粒子数Nrff定义如下: [0033] wherein, the effective number of particles Nrff defined as follows:

_ Kff=I^。 _ Kff = I ^. /-1 /-1

[0035] 其中,所述传感器节点及簇头节点装配有GPS系统,目标定位跟踪系统采用GPS授时的方式实现网络的时间同步。 [0035] wherein said sensor nodes and cluster head node is equipped with a GPS system, the target location and tracking system uses GPS Timing achieve network are time synchronized.

[0036] 本发明提供的一种目标定位跟踪方法,包括如下步骤: [0036] A target tracking method of the present invention to provide positioning, comprising the steps of:

[0037] (1)完成系统配置,包括传感器节点、簇头节点和指控系统。 [0037] (1) complete system configurations, including sensor nodes, and the cluster head node command and control system.

[0038] (2)完成网络初始化。 [0038] (2) to complete the network initialization.

[0039] (3)传感器节点对所在簇内进行监测,依据采集信号判断是否有目标出现。 [0039] (3) where the sensor nodes within the cluster monitoring, determines whether a signal based on the acquisition target appears.

[0040] (4)当出现目标后,传感器节点对目标的方位进行测量。 [0040] (4) When the target appears, a target azimuth sensor node is measured.

[0041] (5)传感器节点将自己的方位测量值及自身坐标信息发送给本簇内的簇头节点。 [0041] (5) The sensor node sends its own orientation measurement value and the coordinate information to a cluster head in a cluster present.

[0042] (6)簇头节点根据本簇内的传感器节点发送的测量值及其坐标信息,完成目标状态估计,实现目标的定位跟踪,并将定位跟踪结果发送到指控模块;包括以下子步骤: [0042] (6) The cluster head and the measured value of the coordinate information transmitted from the sensor nodes within this cluster, the target state estimation is completed, to achieve the target location and tracking, and transmits the result to the alleged location tracking module; comprises the substeps of :

[0043] (61)初始化模块建立目标运动及观测方程、目标的初始状态分布函数和初始状态的估计方差,并随机产生N个粒子。 [0043] (61) and the initialization module establishment of the target observation equation, the variance of the initial estimate of the target state and initial state distribution function, and generate N random particles.

[0044] (62)粒子滤波模块利用状态方程对每个粒子状态进行基于扩展卡尔曼滤波的预测、更新。 [0044] (62) Particle filtering module for each particle based on the predicted state extended Kalman filter is updated using the equation of state.

[0045] (63)粒子权值计算模块计算粒子权值。 [0045] The calculated weight particles (63) of the particle weight value calculation module.

[0046] (64)重采样判断模块根据有效粒子数Nrff判断是否要重采样,若有效粒子数Nrff 小于预先设定值,则进行重采样,根据重要性密度函数重新采样N个粒子并分配权值。 [0046] (64) resampling module determines whether the number of particles based on the effective resampling Nrff determination, if the effective number of particles Nrff less than the preset value, the re-sampling and re-sampling the N particles in accordance with the importance assigned a weight density function value.

[0047] (65)估计目标状态模块根据粒子的权值计算估计目标状态,随观测时刻的增加, 返回步骤(6¾进行迭代算法。 [0047] (65) a target state estimation module calculates an estimated target state according to the particle weight, the observation time increases, returning to step (6¾ iterative algorithm.

[0048] (7)指控模块收集并显示各簇头节点的定位跟踪结果。 [0048] (7) charges collect and display module of each cluster head positioning tracking results.

[0049] 其中,所述步骤(¾包括以下子步骤: [0049] wherein said step (¾ comprises the substeps of:

[0050] (21)传感器节点完成对自己的编号,唯一标示自己的身份。 [0050] (21) sensor nodes complete their number to uniquely identify their own identity.

[0051] (22)传感器节点借助GPS系统实现时间同步。 [0051] (22) by means of a sensor node GPS system time synchronization.

[0052] (23)传感器节点借助GPS系统实现自身定位。 [0052] (23) by means of a sensor node itself GPS positioning system implementation.

[0053] (24)依据传感器节点的通信距离对网络进行簇划分,选择簇头节点。 [0053] (24) based clustering in the network communication distance sensor nodes choose the cluster head node.

[0054] 其中,所述步骤(61)包括以下子步骤:[0055] (611)建立目标运动及观测模型模块建立目标的一阶马尔可夫系统方程及目标观测的观测模型: [0056] [0054] wherein, said step (61) comprises the substeps of: [0055] (611) and establishment of the target observation object model module to establish a first-order Markov equation and the observation target system observation model: [0056]

Figure CN101644758BD00081

[0057] 其中:XT = [xT,vx,yT,Vy]表示目标的运动状态,xT,yT分别表示目标在X轴及Y轴的坐标值,Vx, Vy分别表示目标在在X轴及Y轴的绝对速度分量; [0057] wherein: XT = [xT, vx, yT, Vy] represents the target state, xT, yT represent a target coordinate value X-axis and Y-axis, Vx, Vy denote target in the X-axis and Y component of the absolute speed of the shaft;

Figure CN101644758BD00082

, 表示传感器节点i分别在X轴及Y轴的坐标,N表示观测到目标的传感器节点个数;θ i表示目标与传感器节点的相对方位叫表示第i个传感器节点的测量噪声。 , I represents each sensor node in the X-axis and Y-axis coordinate is, N represents the number of sensor nodes to the observed object; [theta] i represents the relative bearing sensor target node name indicates the i-th measurement noise sensor nodes.

[0058] (612)建立目标初始化状态及估计方差模块建立目标的初始状态概率分布函数及估计方差。 [0058] (612) establishing targets and estimated variance module initialization state goal of establishing an initial state probability distribution function and the estimation variance.

[0059] 所述初始状态概率分布函数为:Xk+1 = FkXk+GkAk。 [0059] The initial state probability distribution function is: Xk + 1 = FkXk + GkAk.

[0060] 其中:Ak= [ax,k, 系统的处理噪声,即由于目标运动环境的不确定性导致分别在X和Y轴引起的加速度噪声,Ak〜N(0,Rk); [0060] wherein: processing the noise Ak = [ax, k, of the system, i.e., the target due to the uncertainty caused by movement in the environment causes each of the X and Y axes acceleration noise, Ak~N (0, Rk);

[0061] [0061]

Figure CN101644758BD00083

[0062] (613)随机粒子产生模块从初始状态分布ρ (Xtl)中随机抽取N个初始粒子{χΛ i =1,2···Ν},与初始粒子对应的协方差矩阵为:{ΡΛ i = 1,2…N}。 [0062] (613) random random particle generation module from the initial state distribution ρ (Xtl) in N initial particles {χΛ i = 1,2 ··· Ν}, corresponding to the initial particle covariance matrix: {ΡΛ i = 1,2 ... N}.

[0063] 其中,所述步骤(6¾包括以下子步骤: [0063] wherein said step (6¾ comprises the substeps of:

[0064] (621)预测模块利用状态方程预测k时刻的目标的状态及预测协方差矩阵H似其中 [0064] (621) using the predicted target state prediction module at time k and the predictive equation of state covariance matrix H wherein like

Figure CN101644758BD00084

[0065] (622)更新模块利用k时刻的测量值对粒子预测结果及估计协方差矩阵进行更新,得到k时刻更新值及协方差矩阵Xki和ΡΛ [0065] (622) using the measured value updating module particle at time k and the estimated prediction results covariance matrix is ​​updated, the updated value obtained at time k and a covariance matrix and ΡΛ Xki

[0066]其中- [0066] wherein -

Figure CN101644758BD00085

[0067] [0067]

[0068] 其中,所述步骤(6¾中,粒子权值计算模块根据贝叶斯及粒子滤波理论计算粒子权值: [0068] wherein, in the step (6¾ in particle weight value calculation module calculates the Bayesian theory particle and particle filter weights:

[0069] [0069]

Figure CN101644758BD00086

[0070] 其中:(1(¾/¾+/, z1:k)是重要性密度函数,p(zk/xkJ)是观测似然函数,P(XkAk^1) 是系统一阶马尔可夫过程的转移概率。 [0070] wherein: (1 (¾ / ¾ + /, z1: ​​k) the importance of the density function, p (zk / xkJ) an observation likelihood function, P (XkAk ^ 1) is a first-order Markov process system the transition probability.

[0071] 本发明的优点在于: [0071] The advantage of the present invention:

[0072] 1、本发明提供的目标定位跟踪系统及方法采用一种结合卡尔曼和粒子滤波的方法实现目标的被动定位,其在实现高精度定位的基础上,运算速度远远低于粒子滤波方法, 具有很高的有效性、精确性及可行性。 [0072] 1, target location and tracking system and method of the present invention provides a method of using a particle filter and a Kalman Passive targeting, which is based on high precision positioning on the operating speed far below the particle filter method with high effectiveness, accuracy and feasibility. [0073] 2、本发明提供的目标定位跟踪系统及方法采用基于传感器网络的系统配置实现目标的运动分析,克服了单节点的可观测性原理的局限。 [0073] 2, the target location and tracking system and method of the present invention provides the use of limited sensor network-based system configured to implement a target motion analysis, to overcome the single-node principle of observability.

[0074] 3、本发明提供的目标定位跟踪系统及方法在进行目标的定位应用中,对系统的配置借鉴无线传感器网络的技术特点,采用多传感器基于对目标方位的观测最终实现对目标的定位跟踪,该网络配置可以克服传统机载或舰载单站系统在观测期间必须进行机动的约束,当采用多传感器测量的目标方位信息实现目标协同定位,不需要观测平台的机动,提高了目标定位的灵活性,而且采用该系统配置实现可以大大增加目标监测定位的区域面积, 避免存在定位盲区的不足。 [0074] 3, the target location and tracking system and method of the present invention to provide a positioning is performed in the target application, the configuration of the system learn the characteristics of the wireless sensor network technology, the use of multi-sensor target orientation based on the observation of the target is ultimately positioned tracking, the network configuration can overcome the traditional airborne or marine single station system must be constrained during the observation maneuver, when the multi-sensor measurement target orientation information goals co-located, the motor does not require the observation platform, improved targeting flexibility, and the use of the system configurations can greatly increase the area of ​​monitoring and location of the target, to avoid the shortcomings of locating blind spots.

附图说明 [0075] 图 1是本发明目标定位跟踪系统配置示意图;[0076] 图 2是本发明传感器节点设计功能模块图;[0077] 图 3是本发明传感器节点功能划分图;[0078] 图 4是本发明目标定位跟踪流程图;[0079] 图 5是本发明定位系统初始化流程图;[0080] 图 6是本发明EKF-PF算法流程图;[0081] 图 7是本发明实施例中传感器节点与目标轨迹跟踪的总体示意图;[0082] 图 8是本发明实施例中目标跟踪的局部放大结果比较示意图;[0083] 图 9是本发明实施例中对目标X轴和Y轴坐标的估计结果比较图;[0084] 图 10是本发明实施例中对目标X轴和Y轴方向速度的估计结果比较图。 BRIEF DESCRIPTION [0075] The object of the present invention, FIG. 1 is a schematic configuration diagram of the location tracking system; [0076] of the present invention, FIG 2 is a functional block diagram of a sensor node design; [0077] FIG. 3 is a functional division FIG sensor nodes invention; [0078] FIG 4 is a flowchart of the target location and tracking of the present invention; [0079] FIG. 5 is a flowchart showing a positioning system according to the present invention is initialized; [0080] FIG. 6 is a flowchart EKF-PF algorithm of the present invention; [0081] FIG. 7 is an embodiment of the present invention the overall schematic diagram of sensor nodes and the target trajectory tracking; [0082] FIG. 8 is a partially enlarged results of Examples target tracking embodiment of the present invention, comparative schematic; [0083] FIG. 9 is an embodiment of the target X-axis and Y-axis coordinates embodiment of the present invention FIG estimation result of the comparison; [0084] FIG. 10 is a comparison example of FIG estimation result of the target X-axis and Y-axis direction of the velocity embodiment of the present invention.

具体实施方式 Detailed ways

[0085] 下面结合附图及一个具体实施例对分发明做详细说明。 [0085] The following drawings and a specific embodiment of the invention, a detailed explanation binding points.

[0086] 本实施例假设对水下一航行器目标进行定位与跟踪,本实施例的目标定位跟踪系统,如图1所示,由多个簇定位模块及一个指控系统组成;每个簇定位模块包括两个传感器节点和一个簇头节点。 [0086] The present embodiment assumes that the water of the next target location and aircraft tracking, targeting embodiment of the tracking system of the present embodiment, as shown, is positioned by a plurality of clusters and a command and control system composed of modules 1; each cluster positioned module comprises two sensor nodes and a cluster head node. 每个簇头节点进一步包括:一初始化模块、一粒子滤波模块、一粒子权值计算模块、一重采样判断模块及一估计目标状态模块。 Each cluster head further comprising: an initialization module, a filter module particle, a particle weight calculation module, a determination module and a resampling target state estimation module.

[0087] 传感器节点静态部署,也可以进行小范围的移动。 [0087] static sensor nodes deployed, may also be a small range of movement. 其位置坐标可通过传感器节点利用自带的GPS装置完成自定位,也可以通过几个已知坐标的锚节点依据一定的自定位协议完成自定位;每个传感器节点需具有测向功能,这个功能可通过采用阵列信号处理手段实现目标的方位估计;为了进行数据的交互和组网功能,传感器节点需具有无线通信功能; 簇头节点负责本小区域内目标定位跟踪,它可做为本小区的网关节点看待;簇头节点存储有该区域内所有传感器节点的位置坐标;并负责收集该簇内传感器节点所测量的目标方位信息;利用改进的快速EKF-PF算法实现对目标的定位跟踪;指控系统负责收集各个区域内簇头节点定位跟踪结果,在终端对定位结果进行显示,以便让定位跟踪结果用于不同的应用需求。 The position coordinates can be done by using self-positioning sensor nodes own GPS device, the positioning can also be done from the anchor node by positioning protocol according to a certain number from the known coordinates; each sensor node needs to have a function to measure, this function can be estimated by using the orientation of the array signal processing means to achieve; networking capabilities to interact with the data, sensor nodes need to have a wireless communication function; cluster head node is responsible for the tracking cell targeting domain, it can do the gateway cell present look node; cluster head node stores the position coordinates of all the sensor nodes in the region; and is responsible for collecting information in the target position sensor node cluster measured; the improved EKF-PF fast algorithm for locating targets tracking; command and control system responsible for collecting the various regions within the cluster head node location tracking results, the positioning result can be displayed on the terminal in order to make locating and tracking the results for different application requirements. 其中簇头节点可以具有测向功能,也可以不具有该功能,但是其必须具有无线通信功能。 Wherein the cluster-head node may have a function to measure, may not have this feature, but it must have a wireless communication function. 簇头节点可以事先规定,也可以在一个簇内动态选择。 Cluster head node can be specified in advance or may be dynamically selected in a cluster.

[0088] 对于整个定位系统的同步问题,采用GPS授时的方式实现网络的时间同步。 [0088] For the synchronization of the whole positioning system, GPS Time by way time synchronization network. 所以本系统内节点需装配有GPS系统。 This node within the system so equipped with GPS systems. [0089] 对于定位系统的网络互连,采用无线通信方式,对于一个簇内传感器节点之间没有通信链路,传感器节点之间无需链接进行数据交换。 [0089] For the interconnection network positioning system, wireless communication, not for a communication link between the sensor nodes within a cluster, data exchange between the sensor nodes without links. 本系统存在三种链路链接方式,第一就是传感器节点与负责该区域内的簇头节点之间的通信链路,负责将自己测量的方位信息发送给簇头节点实现Era-PF定位跟踪算法;第二是簇头节点与指控系统之间的通信链路, 簇头节点负责将定位跟踪结果发送给指控系统进行显示;第三类是各个区域簇头节点之间的通信链路,随着目标的运动,它进入不同的区域,则前一个区域的簇头节点将最终的定位结果发送给此时的簇头节点做为该簇内目标定位跟踪的初始结果。 There are three links linking the system of the present embodiment, the first is the communication link between the sensor nodes and cluster head node is responsible for this region, is responsible for sending the measured position information to their cluster head node implements Era-PF Location Tracking ; a second communication link between the cluster head node command and control system, the cluster head node is responsible for positioning the tracking results to the command and control system for display; third category is the communication link between the various regions of the cluster head node, as moving targets, which enter the different regions, the cluster head transmits before a final area of ​​location result to the case of the cluster head in the cluster as the initial results of the tracking target location.

[0090] 图2给出了本发明中涉及到的传感器节点及簇头节点的设计模块,主要有四个模块组成。 [0090] FIG. 2 shows the present invention relates to a sensor node and cluster head design module, has four main modules. 供电模块完成对整个节点的电能供给;处理模块是节点的计算单元,采用商业的DSP芯片,负责将传感器所采集数据进行分析和处理,并进行数据打包;通信模块,发送打包数据,同时接收其它节点发送的数据包与通信模块一起完成节点的无线通信任务。 The power supply module to complete the power supply to the entire node; processing module is to calculate cell node, using commercial DSP chip, is responsible for the sensor the collected data are analyzed and processed, and the data package; a communication module, transmits the packed data, while receiving other complete wireless communication with the task node data packets sent by the node with the communication module. 对于传感器节点来说,其通信模块器件包括无线射频通信系统。 For the sensor node, the device includes a wireless communication module which radio communications systems. 传感器模块主要完成对目标信号数据进行采集。 A sensor module to complete data acquisition of the target signal.

[0091] 基于传感器节点的在定位系统中的应用,在网络七层协议的基础上对传感器节点数据处理功能进行了简单的划分,主要分为以下三个部分:通信系统;支撑系统;应用系统。 [0091] Based on the application of the sensor node in the positioning system, the seven network protocol based on the sensor data processing node a simple division, is divided into three parts: the communication system; support system; application . 如图3所示,对于通信系统主要由物理层,数据链层,网络层和传输层组成,负责完成节点的正常组网功能。 As shown in FIG. 3, the communication system is mainly composed of a physical layer, data link layer, network layer and transport layer, is responsible for the completion of the normal function of network nodes. 支撑系统包括节点同步与自定位,传感器节点可通过GPS装置完成传感器节点的时间同步与节点自身的定位技术。 The support system includes a node is synchronized with self-location, to be completed by the time the sensor nodes sensor node is synchronized with node itself by GPS positioning technology means. 应用系统主要在基于传感器采集数据的基础上负责完成传感器节点的测向功能,可借鉴阵列信号处理中的相关技术完成。 Application of the system is responsible for data collection based on the sensor based on the sensing function of the sensor nodes is completed, the array signal processing can learn related techniques.

[0092] 利用上述目标定位跟踪系统对目标进行定位跟踪的方法,如图4所示,包括以下步骤: [0092] positioned on the target track using the targeting methods of the tracking system shown in Figure 4, comprising the steps of:

[0093] (1)完成系统配置,包括传感器节点、簇头节点和指控系统。 [0093] (1) complete system configurations, including sensor nodes, and the cluster head node command and control system.

[0094] (2)完成网络初始化,如图5所示,包括以下子步骤: [0094] (2) to complete the network initialization, 5, comprises the substeps of:

[0095] (21)传感器节点完成对自己的编号,唯一标示自己的身份。 [0095] (21) sensor nodes complete their number to uniquely identify their own identity.

[0096] (22)传感器节点借助GPS系统实现时间同步。 [0096] (22) by means of a sensor node GPS system time synchronization.

[0097] (23)传感器节点借助GPS系统实现自身定位。 [0097] (23) by means of a sensor node itself GPS positioning system implementation.

[0098] (24)依据传感器节点的通信距离对网络进行簇划分,选择簇头节点。 [0098] (24) based clustering in the network communication distance sensor nodes choose the cluster head node. 此步骤可以人为指定簇头,也可以随机在簇内选择簇头。 This step can be manually specified cluster head, the cluster head may be randomly selected within a cluster.

[0099] (3)传感器节点对所在簇内进行监测,依据采集信号判断是否有目标出现。 [0099] (3) where the sensor nodes within the cluster monitoring, determines whether a signal based on the acquisition target appears.

[0100] (4)当出现目标后,传感器节点对目标的方位进行测量。 [0100] (4) When the target appears, a target azimuth sensor node is measured.

[0101] (5)传感器节点将自己的方位测量值及自身坐标信息发送给本簇内的簇头节点。 [0101] (5) The sensor node sends its own orientation measurement value and the coordinate information to a cluster head in a cluster present.

[0102] (6)簇头节点根据本簇内的传感器节点发送的测量值及其坐标信息,完成目标状态估计,实现目标的定位跟踪,并将定位跟踪结果通过无线通信发送到指控模块;如图6所示,包括以下子步骤: [0102] (6) The cluster head and the measured value of the coordinate information transmitted from the sensor nodes within this cluster, the target state estimation is completed, to achieve the target location and tracking, and transmits the result to the alleged location tracking module by wireless communication; as As shown in FIG. 6, comprising the sub-steps of:

[0103] (61)初始化模块建立目标运动及观测方程、目标的初始状态分布函数和初始状态的估计方差,并随机产生N个粒子。 [0103] (61) and the initialization module establishment of the target observation equation, the variance of the initial estimate of the target state and initial state distribution function, and generate N random particles.

[0104] (611)建立目标运动及观测模型模块建立目标的一阶马尔可夫系统方程及目标观测的观测模型:[0105] [0104] (611) and establishment of the target observation object model module to establish a first-order Markov equation and the observation target system observation model: [0105]

Figure CN101644758BD00111

[0106] 其中: [0106] wherein:

Figure CN101644758BD00112

表示目标的运动状态,xT,yT分别表示目标在X轴及Y轴的坐标值,Vx, Vy分别表示目标在在X轴及Y轴的绝对速度分量;假设目标与水下节点在同一个平面,节点处于静止状态, Represents the target moving state, xT, yT represent target values ​​of the X coordinate axis and the Y-axis, Vx, Vy denote the absolute velocity component of the target in the X-axis and Y-axis in; assumed underwater target node in the same plane , the node in a stationary state,

Figure CN101644758BD00113

'i = 1,2... N表示传感器节点i分 'I = 1,2 ... N denotes a sensor node points i

别在X轴及Y轴的坐标,N表示观测到目标的传感器节点个数;θ i表示目标与传感器节点的相对方位;ni表示第i个传感器节点的测量噪声。 Do not coordinate X-axis and Y-axis, N is the number of sensor nodes to the observed object; [theta] i represents the relative orientation of the target and sensor node; Ni represents the i-th measurement noise sensor nodes. 系统在k时刻的观测模型可以用矩阵形式表示: The system observation model at time k can be expressed in matrix form:

[0107] [0107]

Figure CN101644758BD00114

[0108] 其中: [0108] wherein:

Figure CN101644758BD00115
Figure CN101644758BD00116

[0112] 假定噪声为零均值高斯随机噪声,则 [0112] assumed to be zero mean Gaussian noise is random noise, then

Figure CN101644758BD00117

(4为噪声协方差矩阵,且 (4 noise covariance matrix, and

Figure CN101644758BD00118

[0113] (612)建立目标初始化状态及估计方差模块建立目标的初始状态概率分布函数及估计方差。 [0113] (612) establishing targets and estimated variance module initialization state goal of establishing an initial state probability distribution function and the estimation variance.

[0114] 采用白噪声加速度模型(White Noise Acceleration, WNA)来描述目标的弱机动性。 [0114] The white noise acceleration model (White Noise Acceleration, WNA) described weak mobility target. 则所述初始状态概率分布函数为: Then the initial state probability distribution function:

Figure CN101644758BD00119

[0115] 其中: [0115] wherein:

Figure CN101644758BD001110

, 系统的处理噪声,即由于目标运动环境的不确定性导致分别在X和Y轴引起的加速度噪声,Ak〜N(0,Rk); The processing system noise, i.e., object motion due to the uncertainty caused in the environment causes each of the X and Y axes acceleration noise, Ak~N (0, Rk);

Figure CN101644758BD001111

[0117] (613)随机粒子产生模块从初始状态分布ρ (Xtl)中随机抽取N个初始粒子{χΛ i =1,2···Ν},与初始粒子对应的协方差矩阵为:{ΡΛ i = 1,2…N}。 [0117] (613) random random particle generation module from the initial state distribution ρ (Xtl) in N initial particles {χΛ i = 1,2 ··· Ν}, corresponding to the initial particle covariance matrix: {ΡΛ i = 1,2 ... N}.

[0118] 通过上述测量方程和状态方程可以看出,基于水下多传感器方位TMA问题的状态方程是线性的,而测量方程是非线性的,而且是时变的,因此实际是一个时变非线性滤波问题。 [0118] can be seen from the above state equation and the measurement equation, based on the state Underwater Multisensor orientation TMA problem equation is linear, and the measurement equations are nonlinear and time-varying, thus a time varying nonlinear actually filtering problem.

[0119] 针对传统的粒子滤波算法对上述问题的求解,影响粒子滤波估计效果的重要因素是重要性密度函数的选择,而对于最优重要性密度函数是后验概率密度,所以自然而然想到将扩展卡尔曼滤波方法与粒子滤波方法相结合,在k时刻先验粒子通过EKF滤波更新后, 其值更加接近于后验概率,所以与PF相比,当达到同一性能所需的粒子数要少于PF算法, 而且该算法对PF中粒子耗尽问题也有一定的抑制作用。 [0119] For solving the traditional particle filter algorithm to the problem, an important factor to estimate the effect of the particle filter is to choose the importance density function, and for optimal importance density function is the posterior probability density, so naturally think of to expand Kalman filtering and combining particle filtering, particle at time k after a priori by EKF filter update, which is closer to the value of posterior probability, as compared with the PF, when reaching the desired number of particles to be less than the same properties PF algorithm, and the algorithm PF particle depletion problem also has a certain extent. [0120] (62)粒子滤波模块利用状态方程对每个粒子状态进行基于扩展卡尔曼滤波的预测、更新。 [0120] (62) Particle filtering module for each particle based on the predicted state extended Kalman filter is updated using the equation of state.

[0121] (621)预测模块利用状态方程预测k时刻的目标的状态劣/h及预测协方差矩阵 [0121] (621) using the target state prediction module prediction time k deterioration state equation / h and predicted covariance matrix

Figure CN101644758BD00121

[0122] (622)更新模块利用k时刻的测量值对粒子预测结果及估计协方差矩阵进行更新,得到k时刻更新值及协方差矩阵Xki和ΡΛ [0122] (622) using the measured value updating module particle at time k and the estimated prediction results covariance matrix is ​​updated, the updated value obtained at time k and a covariance matrix and ΡΛ Xki

[0123]其中: [0123] wherein:

Figure CN101644758BD00122

[0124] [0124]

Figure CN101644758BD00123

[0125] (63)粒子权值计算模块计算粒子权值。 [0125] calculating the particle weight value (63) the particle weight value calculation module.

[0126] 粒子权值计算模块根据贝叶斯及粒子滤波理论计算粒子权值: [0126] the particle weight value calculation module calculates Bayesian theoretical particle and particle filter weights:

[0127] [0127]

Figure CN101644758BD00124

[0128] 其中:(1(¾/¾+/, z1:k)是重要性密度函数,p(zk/xkJ)是观测似然函数,P(XkAk^1) 是系统一阶马尔可夫过程的转移概率。 [0128] wherein: (1 (¾ / ¾ + /, z1: ​​k) the importance of the density function, p (zk / xkJ) an observation likelihood function, P (XkAk ^ 1) is a first-order Markov process system the transition probability.

[0129] (64)重采样判断模块根据有效粒子数Nrff判断是否要重采样,若有效粒子数Nrff 小于预先设定值,则进行重采样,根据重要性密度函数重新采样N个粒子并分配权值。 [0129] (64) resampling module determines whether the number of particles based on the effective resampling Nrff determination, if the effective number of particles Nrff less than the preset value, the re-sampling and re-sampling the N particles in accordance with the importance assigned a weight density function value.

[0130] (65)估计目标状态模块根据粒子的权值计算估计目标状态,随观测时刻的增加, 返回步骤(6¾进行迭代算法。 [0130] (65) a target state estimation module calculates an estimated target state according to the particle weight, the observation time increases, returning to step (6¾ iterative algorithm.

[0131] (7)指控模块收集并显示各簇头节点的定位跟踪结果。 [0131] (7) charges collect and display module of each cluster head positioning tracking results.

[0132] 图7至图10是本实施例中采用EKF-PF算法实现目标定位的计算机仿真结果,为了本发明与传统EKF和PF方法比较,仿真过程对这两种定位方法也进行了仿真试验。 [0132] FIGS. 7 to 10 are present embodiment PF-EKF algorithm embodiment implemented using computer simulation target location, for comparison with the traditional EKF and the PF method of the present invention, these two positioning simulation process simulation test methods . 图7 和图8是跟踪的结果,图8是对图7局部的方法,这样更加直观的看出目标跟踪的效果。 7 and FIG. 8 is a result of the tracking, and FIG. 8 is a partial method of Figure 7, so that more intuitive to see the effect of target tracking. 从图中可以看出,本发明的跟踪效果比EKF方法要好,和PF方法相比其效果相当,但是其计算时间远远要小于PF方法。 As can be seen from the figure, the tracking effect of the present invention is better than the EKF method, compared to the PF method and its results have been very, but the calculation time is much less than the PF method. 图9分别给出了目标在X轴和Y轴坐标的估计值与真实值的对比结果,并接给出了三种不同方法下的误差曲线;图10则是对目标在X轴和Y轴速度的估计值,下表列出了本发明方法的运行时间,可以看出在达到同样的定位精度的情况下,本发明所采用的方法远远低于同样精度的PF方法。 Figure 9 shows the comparative results in the target value of the estimated value and the true X-axis and Y-axis coordinates, and the error curve then gives the three different methods; FIG. 10 is a target in the X-axis and Y-axis the estimated value of the speed, the following table lists the running time of the method of the present invention can be seen in the case to achieve the same positioning accuracy, the method employed in the present invention is much lower than the PF method is the same precision. 通过以上实施例的仿真结果,可以看出本发明方法的有效性和精确性,说明该方法在实际应用的可行性。 Simulation results according to the above embodiment, it can be seen validity and accuracy of the method of the present invention, the feasibility of the method described in the practical application.

[0133] [0133]

Figure CN101644758BD00125

Claims (10)

1. 一种目标定位跟踪系统,包括:多个簇定位模块及一指控模块; 所述簇定位模块包括:多个传感器节点及一簇头节点;所述传感器节点,用于对其所在簇内区域进行监测,当有目标出现,则对目标的方位进行测量,并将该测量值与自身坐标信息发送到本簇内的簇头节点;所述簇头节点,用于根据本簇内的传感器节点发送的测量值及其坐标信息,完成目标状态估计,实现目标的定位跟踪,并将定位跟踪结果发送到指控模块; 所述指控模块,用于收集并显示各簇头节点的定位跟踪结果; 其特征在于,所述簇头节点进一步包括:一初始化模块,用于建立目标运动及观测方程、目标的初始状态分布函数和初始状态的估计方差,并随机产生N个粒子;一粒子滤波模块,用于利用状态方程对每个粒子状态进行基于扩展卡尔曼滤波的预测、更新;一粒子权值计 A target position tracking system, comprising: a positioning module and a plurality of clusters alleged module; locating the cluster module comprising: a plurality of sensor nodes and a cluster head node; the sensor node, located within the cluster for its monitoring area, when the target appears, then the azimuth of the target is measured, and the measured value is transmitted to the coordinate information itself within the cluster head of the present cluster; the cluster head node for a sensor according to the present cluster measured value and the coordinate information sent from the node to complete the estimation target state, to achieve the target location and tracking, and transmits the result to the alleged location tracking module; charges the module, for collecting and displaying the location tracking result of each cluster head node; wherein the cluster-head node further comprises: an initialization module for establishing a target motion and the observation equation, the variance of the initial estimate the target state and initial state distribution function, and generating N random particles; a particle filter module, using the state equation for each particle based on the predicted state extended Kalman filter update; a particle weight basis 算模块,用于计算粒子权值;一重采样判断模块,用于根据有效粒子数Nrff判断是否要重采样,若有效粒子数Nrff小于预先设定值,则进行重采样,根据重要性密度函数重新采样N个粒子并分配权值;及一估计目标状态模块,用于根据粒子的权值计算估计目标状态; 其中,所述有效粒子数Nrff定义如下: Calculation means for calculating the particle weight; determining a resampling module for resampling in accordance with whether the number of effective particles Nrff determination, if the effective number of particles Nrff less than the preset value, the re-sampling, according to the importance of re-density function samples N particles and assigning a weight value; and a target state estimation module configured to calculate an estimated weight of the particle in accordance with the target state value; wherein said effective number of particles Nrff defined as follows:
Figure CN101644758BC00021
其中,N表示选用粒子数目,《表示第k次观察第i个粒子所对应的权值。 Wherein, N represents the number of selected particle "denotes the k-th observation value of the weight corresponding to the i-th particle.
2.根据权利要求1所述的目标定位跟踪系统,其特征在于,所述初始化模块进一步包括:一建立目标运动及观测模型模块、一建立目标初始化状态及估计方差模块及一随机粒子产生模块;所述建立目标运动及观测模型模块,用于建立目标的一阶马尔可夫系统方程及目标观测的观测模型: 2. The tracking system of targeting according to claim 1, wherein said initialization module further comprises: establishing a target motion and the observation model module, and a goal of establishing an initialization state estimation module, and a random variance of particle generation module; the establishment of the target model and the observation means for observing a target model order Markov system equation and the observation target:
Figure CN101644758BC00022
其中:χτ = [χτ, Vx, yT, Vy]表示目标的运动状态,xT, yT分别表示目标在X轴及Y轴的坐标值,vx,Vy分别表示目标在在X轴及Y轴的绝对速度分量;&=[<,<] i = 1,2…N,表示传感器节点i分别在X轴及Y轴的坐标,N表示观测到目标的传感器节点个数;θ i表示目标与传感器节点的相对方位叫表示第i个传感器节点的测量噪声;所述建立目标初始化状态及估计方差模块,用于建立目标的初始状态概率分布函数及估计方差;所述初始状态概率分布函数为:Xk+1 = FkXk+GkAk ;其中:Ak= [ax,k, ay,k]T为系统的处理噪声,即由于目标运动环境的不确定性导致分别在X和Y轴引起的加速度噪声,Ak〜N(0,Rk); Wherein: χτ = [χτ, Vx, yT, Vy] represents the target state, xT, yT represent a target coordinate value X-axis and Y-axis, vx, Vy represent the target in the X-axis and Y-axis of the absolute velocity component; & = [<, <] i = 1,2 ... N, represents a sensor node i are coordinates of the X-axis and Y-axis, N is the number of sensor nodes to the observed object; [theta] i represents the target and sensor node showing relative orientation measurement noise is called i-th sensor node; object initialization state and establishing the variance estimation module, for establishing the initial state probability distribution function of the target and the estimated variance; the initial state probability distribution function is: Xk + 1 = FkXk + GkAk; wherein: Ak = [ax, k, ay, k] T is the noise of the system processing, i.e., due to the uncertainty target motion acceleration noise environment results in the X and Y axes, respectively, due to, Ak~N (0, Rk);
Figure CN101644758BC00023
Figure CN101644758BC00031
所述随机粒子产生模块,用于从初始状态分布P(Xtl)中随机抽取N个初始粒子K,/= I,2…#},与初始粒子对应的协方差矩阵为:{0 = l,2-iV}。 The random particle generation means for randomly N initial particles K P (Xtl) from an initial state distribution, / = I, 2 ... #}, corresponding to the initial particle covariance matrix: {0 = l, 2-iV}.
3.根据权利要求1所述的目标定位跟踪系统,其特征在于,所述粒子滤波模块进一步包括:一预测模块和一更新模块。 3. The position tracking system of the target according to claim 1, wherein the particle filtering module further comprises: a prediction module and an updating module.
4.根据权利要求1所述的目标定位跟踪系统,其特征在于,所述粒子权值计算模块根据贝叶斯及粒子滤波理论计算粒子权值。 The targeting of the tracking system of claim 1, wherein the particle weight value calculation module calculates a weight according to the particle and the particle filter Bayesian theory.
5.根据权利要求1所述的目标定位跟踪系统,其特征在于,所述传感器节点及簇头节点装配有GPS系统,目标定位跟踪系统采用GPS授时的方式实现网络的时间同步。 The targeting of the tracking system according to claim 1, wherein said sensor nodes and cluster head node is equipped with a GPS system, the target location and tracking system uses GPS Timing achieve network are time synchronized.
6. 一种目标定位跟踪方法,包括如下步骤:(1)完成系统配置,包括传感器节点、簇头节点和指控系统;(2)完成网络初始化;(3)传感器节点对所在簇内进行监测,依据采集信号判断是否有目标出现;(4)当出现目标后,传感器节点对目标的方位进行测量;(5)传感器节点将自己的方位测量值及自身坐标信息发送给本簇内的簇头节点;(6)簇头节点根据本簇内的传感器节点发送的测量值及其坐标信息,完成目标状态估计,实现目标的定位跟踪,并将定位跟踪结果发送到指控模块;包括以下子步骤:(61)初始化模块建立目标运动及观测方程、目标的初始状态分布函数和初始状态的估计方差,并随机产生N个粒子;(62)粒子滤波模块利用状态方程对每个粒子状态进行基于扩展卡尔曼滤波的预测、更新;(63)粒子权值计算模块计算粒子权值;(64)重采样判断模块根据有效 A method of target position tracking, comprising the steps of: (1) complete system configurations, including sensor nodes, and the cluster head node command and control system; (2) the completion of the network initialization; (3) where the sensor nodes within the cluster monitoring, based on collected signal determines whether target occurs; (4) when the target appears, the sensor nodes azimuth of the target is measured; (5) the sensor node transmits its orientation measurement value and its own coordinate information to a cluster head in the present cluster ; (6) the measured values ​​of the cluster head and the coordinate information transmitted from the sensor nodes within this cluster, the target state estimation is completed, to achieve the target location and tracking, and transmits the result to the alleged location tracking module; includes the sub-steps :( 61) initialization module and establishment of the target observation equation, the variance of the initial estimate of the target state and initial state distribution function, and generating N random particles; (62) each particle of the particle filter module based on the state equation of the extended Kalman utilization state predictive filtering, updating; (63) value calculation module calculating the particle weight particle weight; (64) based on the effective resampling determination module 子数Nrff判断是否要重采样,若有效粒子数Nrff小于预先设定值,则进行重采样,根据重要性密度函数重新采样N个粒子并分配权值;(65)估计目标状态模块根据粒子的权值计算估计目标状态,随观测时刻的增加,返回步骤(6¾进行迭代算法;(7)指控模块收集并显示各簇头节点的定位跟踪结果; 所述有效粒子数Nrff定义如下: Determining whether the number of sub-Nrff be resampled, if the effective number of particles Nrff less than the preset value, the re-sampling, resampling N particles according to density function and importance weights assigned; (65) a target state estimation module according particles weight calculating an estimated target state, the observation time increases, returning to step (6¾ iterative algorithm; (7) charges the location tracking module collects and displays the results of the cluster head; Nrff the effective number of particles is defined as follows:
Figure CN101644758BC00032
其中,N表示选用粒子数目,巧表示第k次观察第i个粒子所对应的权值。 Wherein, N represents the number of particles selected, Qiao represents the k-th observation value of the weight corresponding to the i-th particle.
7.根据权利要求6所述的目标定位跟踪方法,其特征在于,所述步骤(¾包括以下子步骤:(21)传感器节点完成对自己的编号,唯一标示自己的身份;(22)传感器节点借助GPS系统实现时间同步;(23)传感器节点借助GPS系统实现自身定位;(24)依据传感器节点的通信距离对网络进行簇划分,选择簇头节点。 The target location tracking method according to claim 6, wherein said step (¾ includes the substeps of: (21) number of sensor nodes complete their own, uniquely identify its own identity; (22) sensor node by means of GPS time synchronization system implementation; (23) by means of a sensor node itself GPS positioning system implementation; (24) based clustering in the network communication distance sensor nodes choose the cluster head node.
8.根据权利要求6所述的目标定位跟踪方法,其特征在于,所述步骤(61)包括以下子步骤:(611)建立目标运动及观测模型模块建立目标的一阶马尔可夫系统方程及目标观测的观测模型: The target location tracking method according to claim 6, wherein said step (61) comprises the substeps of: (611) establishment of the target and observed a first-order Markov model module system of equations establish goals and target observation observation model:
Figure CN101644758BC00041
其中:XT = [xT, vx, yT, vy]表示目标的运动状态,xT, yT分别表示目标在X轴及Y轴的坐标值,vx,Vy分别表示目标在在X轴及Y轴的绝对速度分量;S' =[<,<] i = 1,2…N,表示传感器节点i分别在X轴及Y轴的坐标,N表示观测到目标的传感器节点个数;θ i表示目标与传感器节点的相对方位叫表示第i个传感器节点的测量噪声;(612)建立目标初始化状态及估计方差模块建立目标的初始状态概率分布函数及估计方差;所述初始状态概率分布函数为:Xk+1 = FkXk+GkAk ;其中:Ak= [ax,k, ay,k]T为系统的处理噪声,即由于目标运动环境的不确定性导致分别在X和Y轴引起的加速度噪声,Ak〜N(0,Rk); Wherein: XT = [xT, vx, yT, vy] represents the target state, xT, yT represent a target coordinate value X-axis and Y-axis, vx, Vy represent the target in the X-axis and Y-axis of the absolute velocity component; S '= [<, <] i = 1,2 ... N, represents a sensor node i are coordinates of the X-axis and Y-axis, N is the number of sensor nodes to the observed object; [theta] i represents the target and the sensor nodes represent the relative orientation of the measurement noise is called i-th sensor node; establishing the initial state of the target (612) to establish a target variance estimation module initialization state and a probability distribution function and the estimated variance; the initial state probability distribution function is: Xk + 1 = FkXk + GkAk; wherein: Ak = [ax, k, ay, k] T is the noise of the system processing, i.e., due to the uncertainty target motion acceleration noise environment results in the X and Y axes, respectively, due to, Ak~N ( 0, Rk);
Figure CN101644758BC00042
(613)随机粒子产生模块从初始状态分布P(Xtl)中随机抽取N个初始粒子㈨,/= I,2…#},与初始粒子对应的协方差矩阵为:{0 = ι>··#}。 (613) N random particles generated random initial particles (ix) P (Xtl) in the distribution module from the initial state, / = I, 2 ... #}, corresponding to the covariance matrix of primary particles: ·· {0 = ι> #}.
9.根据权利要求6所述的目标定位跟踪方法,其特征在于,所述步骤(6¾包括以下子步骤:(621)预测模块利用状态方程预测k时刻的目标的状态及预测协方差矩阵;(622)更新模块利用k时刻的测量值对粒子预测结果及估计协方差矩阵进行更新,得到k时刻更新值及协方差矩阵。 9. The method of target position tracking according to claim 6, wherein said step (6¾ includes the substeps of: (621) a target state prediction module uses the predicted state at time k and the prediction covariance matrix equation; ( 622) updating module using the measurement at time k and the estimated prediction results of particle covariance matrix update, the updated value obtained at time k and a covariance matrix.
10.根据权利要求6所述的目标定位跟踪方法,其特征在于,所述步骤(6¾中,粒子权值计算模块根据贝叶斯及粒子滤波理论计算粒子权值。 10. A target tracking method of locating according to claim 6, characterized in that, (6¾ in particle weight value calculation module calculates the Bayesian theory of particle filter and the particle weights step.
CN 200910078474 2009-02-24 2009-02-24 An object localization and tracking system and method CN101644758B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200910078474 CN101644758B (en) 2009-02-24 2009-02-24 An object localization and tracking system and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200910078474 CN101644758B (en) 2009-02-24 2009-02-24 An object localization and tracking system and method

Publications (2)

Publication Number Publication Date
CN101644758A CN101644758A (en) 2010-02-10
CN101644758B true CN101644758B (en) 2011-12-28

Family

ID=41656721

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200910078474 CN101644758B (en) 2009-02-24 2009-02-24 An object localization and tracking system and method

Country Status (1)

Country Link
CN (1) CN101644758B (en)

Families Citing this family (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101819682A (en) * 2010-04-09 2010-09-01 哈尔滨工程大学 Target tracking method based on Markov chain Monte-Carlo particle filtering
CN102568004A (en) * 2011-12-22 2012-07-11 南昌航空大学 Tracking algorithm for high maneuvering targets
CN102542577A (en) * 2011-12-22 2012-07-04 电子科技大学 Particle state evaluation method
CN102830402B (en) * 2012-09-10 2014-09-10 江苏科技大学 Target tracking system and method for underwater sensor network
CN103052128A (en) * 2012-12-20 2013-04-17 华南理工大学 Wireless sensor network-based energy-efficient collaborative scheduling method
CN103152791B (en) * 2013-01-29 2015-08-19 浙江大学 A kind of method for tracking target based on underwater wireless sensor network
CN103645487B (en) * 2013-12-06 2017-04-05 江苏科技大学 Underwater multi-target tracking
CN105023277B (en) * 2014-04-15 2018-10-12 南京理工大学 Kalman's particle filter tracking method based on complicated dynamic scene
CN104050686B (en) * 2014-06-24 2017-12-26 重庆硕奥科技有限公司 A kind of dense space method for tracking target
CN105699977B (en) * 2014-11-25 2017-11-21 中国科学院声学研究所 A kind of tracking for moving frogman
CN104469875B (en) * 2014-11-26 2018-01-12 北京邮电大学 Method for tracking target and its system based on prediction in radio sensing network
CN104880707B (en) * 2014-11-30 2017-09-26 中国科学院沈阳自动化研究所 A kind of Interactive Multiple-Model tracking based on adaptive transition probability matrix
CN105548985B (en) * 2015-12-29 2018-05-04 中国人民解放军海军航空工程学院 Maneuvering target tracking method based on RAV-Jerk models
CN107132504A (en) * 2016-02-29 2017-09-05 富士通株式会社 Location tracking device, method and electronic equipment based on particle filter
CN106123892A (en) * 2016-06-22 2016-11-16 武汉科技大学 A kind of robot localization method based on wireless sensor network Yu earth magnetism map
CN106908762B (en) * 2017-01-12 2019-10-22 浙江工业大学 A kind of more hypothesis UKF method for tracking target for UHF-RFID system

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101251593A (en) 2008-03-31 2008-08-27 中国科学院计算技术研究所;北京邮电大学 Method for tracking target of wireless sensor network

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101251593A (en) 2008-03-31 2008-08-27 中国科学院计算技术研究所;北京邮电大学 Method for tracking target of wireless sensor network

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
何祖军等.基于无味粒子滤波和交互多模型算法的多动机目标跟踪.《江苏科技大学学报(自然科学版)》.2008,第22卷(第6期),48-52.
崔平远.基于卡尔曼/粒子组合滤波器的组合导航方法研究.《系统仿真学报》.2009,第21卷(第1期),220-223.
谭爱国.单站无源定位中的一种简单粒子滤波算法.《湖北民族学院学报(自然科学版)》.2007,第25卷(第4期),411-414.
贺静波等.基于改进粒子滤波的双站无源定位跟踪算法研究.《系统仿真学报》.2008,第20卷(第11期),2825-2827.

Also Published As

Publication number Publication date
CN101644758A (en) 2010-02-10

Similar Documents

Publication Publication Date Title
Wang et al. Target tracking in wireless sensor networks based on the combination of KF and MLE using distance measurements
Shu et al. Gradient-based fingerprinting for indoor localization and tracking
CN102170697B (en) Indoor positioning method and device
Shareef et al. Localization using neural networks in wireless sensor networks
Frampton Acoustic self-localization in a distributed sensor network
CN101965052B (en) Wireless sensing network node positioning method based on optimal beacon set
Zhang et al. A novel distributed sensor positioning system using the dual of target tracking
CN104809326B (en) A kind of asynchronous sensor spatial registration algorithm
CN101655561A (en) Federated Kalman filtering-based method for fusing multilateration data and radar data
CN102209382A (en) Wireless sensor network node positioning method based on received signal strength indicator (RSSI)
CN103379619B (en) A kind of localization method and system
CN107980100A (en) Distributed positioning system and method and self-locating devices
CN101975575A (en) Multi-target tracking method for passive sensor based on particle filtering
CN102096086A (en) Self-adaptive filtering method based on different measuring characteristics of GPS (Global Positioning System)/INS (Inertial Navigation System) integrated navigation system
Cheng et al. Localization in sensor networks with limited number of anchors and clustered placement
Purvis et al. Estimation and optimal configurations for localization using cooperative UAVs
CN102830402B (en) Target tracking system and method for underwater sensor network
Chung et al. Scheduling for distributed sensor networks with single sensor measurement per time step
CN103152824B (en) Positioning method of node in wireless sensor network
Yang et al. Sequence localization algorithm based on 3D voronoi diagram in wireless sensor network
CN102331581A (en) Rapid positioning method of binary TDOA/FDOA satellite-to-earth integration positioning system
Kuang et al. A new distributed localization scheme for wireless sensor networks
CN103729859B (en) A kind of probability nearest neighbor domain multi-object tracking method based on fuzzy clustering
Chen et al. A hybrid prediction method for bridging GPS outages in high-precision POS application
CN103648108B (en) Sensor network distributed consistency object state estimation method

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C14 Granted