摘要
近年来,随着计算机技术、通讯技术的发展,多传感器信息融合技术得到了迅速发展,并成为当前信息领域一个十分活跃的研究热点。多传感器信息融合又称为多传感器数据融合,它使用多个传感器对同一目标进行检测,避免了单个传感器的局限性,从而提供更加全面准确的信息。多传感器信息融合作为多源信息综合处理的一项新技术,能够合成来自某一目标的多源信息,产生比单一信息源更精确的估计。
对于含有未知模型参数和噪声(有色噪声及白噪声)统计的多传感器单输出系统,应用现代时间序列分析方法,基于自回归滑动平均(ARMA)新息模型的递推增广最小二乘法或两段递推增广最小二乘法可在线估计未知模型参数和噪声统计。在按状态分量标量加权线性最小方差最优信息融合准则下,分别对含有未知有色观测噪声或未知有色输入噪声的单输出系统提出了自校正分量解耦信息融合Wiener状态估值器;在按标量加权线性最小方差最优信息融合准则下,分别对含有未知模型参数或未知有色观测噪声的单输出系统提出了ARMA信号自校正信息融合Wiener估值器。证明了它们的收敛性,即若ARMA新息模型的参数估计是一致的,则自校正信息融合Wiener估值器将收敛于当模型参数和噪声方差己知时的最优信息融合Wiener估值器,且自校正信息融合Wiener估值器的精度高于每个局部自校正Wiener估值器。大量跟踪系统的仿真例子说明了其正确性和有效性。
Recently, with the development of computer and communication technology, multisensor information fusion technology develops rapidly and becomes an active research focus in current information domain. Multisensor information fusion is also called multisensor data fusion, which uses multisensor to detect the same object, avoids the limitation of single sensor, so it can offer more general and accurate information. As a new technology for multisource date processing, multisensor information fusion can synthesize multisource date from one object, generate more accurate estimation than single source.
For multisensor single output systems with unknown model parameters and noise (colored and white noise) statistics, using the model time series analysis method, based on recursive extend least squares method or two-stage recursive extend least squares method of the autoregressive average moving innovation model, the unknown model parameters and noise statistics can be obtained on line. Under the linear minimum variance optimal information fusion criterion weighted by scales for state components, self-tuning decoupled fusion Wiener state estimators for the single output system with unknown colored measurement (or input) noise are presented respectively; under the linear minimum variance optimal information fusion criterion weighted by scales, ARMA signal self-tuning fusion Wiener estimators for the single output system with unknown model parameters or measurement noise are presented respectively. The convergence of them is proved, i.e if the parameters estimation is consistent, then self-tuning fusion estimator will converge to the optimal fusion estimator with known model parameters and noise statistics, and the accuracy of self-tuning fusion estimator is higher than that of each local self-tuning estimator. Many simulation examples of tracking system show their effectiveness.
引文
[1] 郁文贤,雍少为,郭桂蓉. 多传感器信息融合技术述评[J]. 国防科技大学学报,1994,16(3):1~8.
[2] 康耀红. 数据融合理论与应用[M]. 西安:西安电子科技大学出版社,1997.
[3] 王璇,李春升,周荫清. 多传感器信息融合技术[J]. 北京航空航天大学学报,1994,20(4):402~406.
[4] 何友,王国宏等. 多传感器信息融合及应用[M]. 北京:电子工业出版社,2000,11,1～5.
[5] 何友,关欣,王国宏.多传感器信息融合研究进展与展望[J].宇航学报,2005,26:524~530.
[6] 王顺奎. 国外信息融合技术概述[J]. 红外与激光技术, 1995,24(1):1~4.
[7] 何友,陆大金,彭金宁. 多级式多传感器信息融合中的状态估计[J]. 电子学报,1999,27(8):60～63.
[8] 盛三元,王建华. 联合卡尔曼滤波在多传感器信息融合中的应用[J]. 雷达与对抗,2002,1:27～33.
[9] 崔平远,黄晓瑞. 基于联合卡尔曼滤波的多传感器信息融合算法及其应用[J]. 电机与控制学报,2001,5(3):72～74.
[10] 潘震中. 多传感器信息融合的谢弗一登普斯特方法[J]. 火力与指挥控制,2000,19(3):12～16.
[11] 石松庆. 信息融合与战术光电子系统[J]. 激光与红外,1997,27(4):195～200.
[12] 周丽莉,胡昌振. 多传感器信息融合技术最新发展及其研究展望[J]. 现代引信,1997,(2):42～46.
[13] 吴秋平,万德钧,王庆. 多传感器信息融合研究现状及展望[J]. 导航,1996,l:16～25.
[14] 曾祥群 Li K K,Chan W L 等. 信息融合技术在故障诊断中的应用[EB/OL].http://www.bjx.com.cn/files/wx/jdg/2002-9/4.htm,2002.
[15] 朱允民. 多传感器分布式统计判决[M]. 北京:科学出版社,2000.
[16] 胡学骏,罗中良. 基于统计理论的多传感器信息融合方法[J]. 传感器技术,2002,8:38～43.
[17] 吕孝乐. 数据融合技术研究及其在船舶组合导航中的应用[D]. 哈尔滨:哈尔滨工程大学,2002.
[18] 马国涛,赵亮,李鹏. 基于 Dempster-Shafer 证据推理的多传感器信息融合技术及其应用[J]. 现代电子技术,2003,19:41～44.
[19] Chao.J,Shao K,Jang L.Uncertain information fusion using belief measure and its application to signal classification[I].IEEE International Conference on Multisensor Fusion and Intelligent Systems,1996,12:151～157.
[20] 权太范. 信息融合神经网络一模糊推理理论与应用[M]. 北京:国防工业出版社,2002.
[21] 李玉榕. 信息融合与智能处理的研究[D]. 杭州:浙江大学,2001.
[22] 郭惠勇. 多传感器信息融合技术的研究与进展[J]. 中国科学基金,2005,1:17～21.
[23] Dasarathy B V.Fuzzy Evidential Reasoning Approach to Target Identity and State Fusion in Multisensor Environments. Optical Engineering, 1997,36(3):669～683.
[24] 潘泉,于昕,程咏梅,张洪才. 信息融合理论的基本方法与进展[J]. 自动化学报,2003,29(4):599～615.
[25] Syrmos Vl, Lewis F L. Robust Eigenvalue Assignment for Generalized Systems[J]. Automatic,1992,38(6):1223-1228.
[26] R.K.SAHA, K.C.CHANG. An Efficient Algorithm for Multisensor Track Fusion[J]. IEEE Transactions on Aerospace and Electronic System Vol.34.NO.1 January 1998.
[27] 周锐,申功勖,房建成. 基于信息融合的目标图像跟踪[J]. 电子学报,1998,26(12):89～91.
[28] 张尧庭,桂劲松. 人工智能中的概率统计方法[M]. 北京:科学出版社,1998.
[29] 袁军,王敏,黄心汉,陈锦江. 智能系统多传感器信息融合研究进展[J]. 控制理论与应用,1994,11(5):513～518.
[30] 王耀南,李树涛. 多传感器信息融合及其应用综述[J]. 控制与决策,2001,16(5):518～522.
[31] 张明路,戈新良,唐智强,刘兴荣. 多传感器信息融合技术研究现状和发展趋势[J]. 河北工业大学学报,2003,32(2).
[32] 张友民,戴冠中,张洪才. 卡尔曼滤波计算方法研究进展[J]. 控制理论与应用,1995,12(5).
[33] Hong Z Q, Hong Y Z, Hong J. Fusion Algorithm of Correlated Local Estimates[J]. Aerospace Science and Technology,2004,8:619~626.
[34] Hashemipour H R, Roy S and A Laub(1998). Decentralized Structures for Parallel Kalman Filtering[J]. IEEE Transactions on Automatic Control,33(1):88~93.
[35] Paik B S, and J H Oh(2000). Gain Fusion Algorithm for Decentralised Parallel Kalman Filters[J]. IEEE Proc.Control theory Appl.147(1):97~103.
[36] Y M Zhu, Z H You, J Zhao, K S Zhang, and X R Li(2001). The Optimality for the Distributed Kalman Filtering Fusion with feedback[J]. Automatica,37,1489~1493.
[37] Carlson N A. Federated Square Root Filter for Decentralized Parallel Processes[J]. IEEE Trans.Aerospace Electronic Systems,1990,AES-26, 517~525.
[38] Carlson N A. Federated Filter for Fault-Tolerant Integated Navigation Systems[J]. Proc.of IEEE PLANS’88,Orlando,FL,1988:110~119.
[39] Loomis PVW, Carlson N A. Berlarducci M P. Common Kalman Filter:Fault-Tolerant Navigation for Next Generation Aircraft[J]. Proc.of theInst.of Navigation Conf.,Santa Barbara,CA,1988.38~45.
[40] NEAL.A.CARLSON. Federated Square Root Filter for Decentralized Parallel Processes. IEEE Transactions on Aerospace and Electronic System Vol.26.NO.3 May 1990.
[41] Kim K H. Development of track to track fusion algorithm. Proceeding of the American Control Conference,Maryland,June1994:1037～1041.
[42] 孙书利,邓自立. 多传感器线性最小方差最优信息融合估计准则[J]. 科学技术与工程,2001,1(5):334～336.
[43] 邓自立,高媛,马建为. 两传感器信息融合最优白噪声反卷积 Wiener 滤波器. 科学技术与工程,2003,3(3):216～218.
[44] 邓自立,高媛,马建为. 两传感器信息融合最优 Kalman 滤波器. 科学技术与工程,2003,3(3):213～215.
[45] 邓自立,祁荣宾. 多传感器信息融合次优稳态 Kalman 滤波器. 中国学术期刊文摘(科技快报),科学技术与工程,2000,6(2):183～184.
[46] Q.Lang Gan,Chris J.Harris.Comparison of Two Measurement Fusion Methods for Kalman-Filter-Based Multisensor Data Fusion, IEEE Transactions on Aerospace and Electronic System Vol.37.NO.1 January 2001.
[47] 邓自立,崔崇信,白敬刚. 基于稳态 Kalman 滤波的两种观测融合方法的功能等价性. 科学技术与工程,2004,7.
[48] Nikoukhah R, Willsky A S and Bernard C L. Kalman Filtering and Riccati Equatiams for Descriptor Systems. IEEE Trans. Automatic Control, 1992,37(9):1325~1341.
[49] 邓自立,郭一新. 现代时间序列分析及其应用[M]. 北京:知识出版社,1998.
[50] 邓自立. 自校正滤波理论及其应用—现代时间序列分析方法[M]. 哈尔滨:哈尔滨工业大学出版社,2003.
[51] Shu-Li Sun,Zi-Li Deng. Mufti-sensor optimal information fusion Kalman filter. Automatica 2004,40:1017～1029.
[52] Shu-Li Sun,Z.L.Deng. Mufti-sensor information fusion White noise filter weighted by sealers based on Kalman predictor. Automatica 2004,40:1447～1453.
[53] 邓自立,马建为,高媛. 两传感器自校正信息融合 Kalman 滤波器. 科学技术与工程,2003,3(4):321～324.
[54] 李云,王欣,邓自立. 多传感器分布式信息融合 Wiener 状态估值器. 科学技术与工程,2005,5(23):1785～1791.
[55] 邓自立. 最优估计理论及其应用—建模、滤波、信息融合估计[M]. 哈尔滨:哈尔滨工业大学出版社,2005.
[56] Deng ZL, Gao Y, Mao L,et al. New approach to information fusion steady-state Kalman filtering. Automatica,2005,41 (10):1695～1707.
[57] 邓自立. 卡尔曼滤波与维纳滤波—现代时间序列分析方法[M]. 哈尔滨:哈尔滨工业大学出版社,2001.
[58] 陈传璋. 数学分析. 上海:上海科学技术出版社,1962.