【作者】 陈玉坤；

【导师】 司锡才；

【作者基本信息】 哈尔滨工程大学 ， 通信与信息系统， 2007， 博士

【摘要】 信息融合技术是多模复合制导武器的关键技术之一,它对改善精确制导武器的性能具有至关重要的意义。本文主要针对多模复合制导导引头技术研究课题中的信息融合技术,论证了信息融合分机的软、硬件设计和实现。根据系统的总体设计要求,其信息融合部分需要进行决策融合、数据关联和数据融合。文中研究了多模导引头信息融合所采用的算法,并在研制出的信息融合分机中运用了上述优选算法。同时,对信息融合的这几个关键技术进行了深入的理论研究。在数据预处理方面,研究了异常数据的识别、修补和时间对准方法。量测数据中野值的剔除方法对跟踪性能有很大影响,针对干扰影响大小的不同,提出了一种对新息在线自适应加权的方法。该方法首先进行野值的判决并剔除,在判决并去掉野值后,对新息乘以一个单调下降的加权函数,根据新息的大小自适应在线调整其权值,改变量测值对滤波估计的修正作用,从而达到提高滤波精度的目的。在决策融合理论方面,主要研究证据理论,侧重于冲突情况下的证据理论处理,研究了证据理论存在的问题以及相应的改进算法。改进算法可以充分利用各传感器信息,并对组合结果进行自适应选择,自适应选择的结果与传感器精度、可靠性以及判决门限值有关。此算法可以解决冲突情况下的目标决策问题。在数据关联方面,研究了最近邻法(NN)、概率数据关联算法(PDA)和联合概率数据关联(JPDA)算法。在杂波环境中,概率数据关联和联合概率数据关联算法是目前十分有用的跟踪算法,但是一旦出现某种干扰或是故障,通过概率数据关联算法得到的滤波值就会偏离真实值很多,造成滤波发散,严重影响性能。针对这一不足,基于概率数据关联算法中的组合新息,提出了修正概率数据关联算法,并进行了对比仿真,仿真结果验证了修正算法的有效性。在融合算法方面,研究了状态向量融合、量测向量融合、离散卡尔曼滤波递推方程和数据融合算法。根据理论和实验结果,提出了一种新的状态融合模型,即修正的测量航迹—航迹融合模型(MMTF),作为三模导引头数据融合的数学模型,并为模型提供了理论依据。基于卡尔曼滤波预测、修正的思想,对雷达、红外传感器数据融合问题进行了研究,提出了雷达、红外传感器数据融合的新方法,仿真和实验结果表明,它可使主动雷达、红外传感器达到较好的数据融合效果。在系统实现方面,根据系统的指标要求,给出了基于高速DSP处理器TMS320C6201和FPGA的整体硬件设计、系统工作过程和软件流程,通过RS—422串口和主机口(HPI口)实现融合分机与被动雷达、主动雷达、红外传感器和弹上计算机的通信,完成了整个设计流程的调试。通过高山对海试验的检验,达到了实时跟踪的效果。更多还原

【Abstract】 Information fusion is one of the key technologies for multimode guidance weapons, which plays an important role in improving precision guidance weapons performance. The dissertation is mainly about the information fusion part of multimode combined guidance seeker technology research, the software and hardware’s design of information fusion is presented. The key technologies of information fusion are decision-making fusion, data association and track fusion. In this dissertation, the information fusion algorithms employed in the multimode seeker are formulated, at the same time the information fusion module is made, which demonstrates those effective algorithms. Meanwhile the detailed theory research is further developed in the dissertation.In data preprocessing, the methods of abnormal value identification and modification and time adjusting are studied. The elimination methods of outliers among the measurements play a crucial part in improving tracking performance. A new way which automatically weightens innovation is proposed according to interference influence. For increasing target tracking accuracy, the method first distinguishes and eliminates outliers, and then innovation multiplies with a descending function. The function weight value is adaptively adjusted according to innovation, and estimation value modified by measurements is changed.In decision-making fusion, this dissertation emphasizes on evidence theory, particularly on evidence theory under conflict. About evidence theory, some existent problems are presented, and the improved algorithm is proposed. The improved algorithm can make good use of multi-senor information; the combined result can be adaptively selected based on the precision and reliability of sensor and judging threshold. The algorithm can solve the problem of target decision fusion under conflict. In data association, nearest neighbor data association (NNDA), probabilistic data association (PDA) and joint probabilistic data association (JPDA) are analyzed. The PDA and JPDA deal with the target tracking perfectly in the clutter environment and have found widespread application in many areas. But if interference appears, the estimate values through PDA will be incorrect and diverge, and the performance will deteriorate dramatically. Based on the federated innovation of the PDA, the modified PDA (MPDA) is proposed. Simulation results demonstrate its effectiveness.In fusion algorithm, the status vector fusion, measurement vector fusion, discrete Kalman equation and track fusion are studied. A new state fusion model is proposed based on the theory and experiment results, then the modified measurement track-to-track fusion (MMTF) is employed as the tri-mode seekers data fusion model. Data fusion of radar and IR is studied based on prediction and correlation of Kalman filter, and then a new data fusion is introduced. The simulation indicates that the algorithm has preferable fusion result by using radar and IR data.In system achievement, according to system guide line, the hardware design, work process and software flow chart based on DSP (TMS320C6201) and FPGA are given. The communications within Passive, Radar, IR and computer are carried out though RS-422 serial port and host-port interface. The whole design process has already been debugged. The real-time result is satisfied through mountain-to-ocean test.更多还原