【作者】 宗华；

【导师】 权太范；

【作者基本信息】 哈尔滨工业大学 ， 信息与通信工程， 2010， 博士

【摘要】 高频地波雷达工作在频率十分拥挤、电磁环境极其复杂的短波段,其探测性能易受电磁环境变化的影响。在复杂电磁环境下,系统探测难以保证目标点迹的连续性,同时较低的方位分辨率导致远距离上检测点迹的高度分散和跳变。因此,在雷达数据处理时,存在难以航迹起始和形成稳定航迹的问题。然而双/多基地高频地波雷达体制是进一步提高单基地高频地波雷达全天候性能、保障在复杂电磁环境下航迹起始和稳定航迹形成的有效途径之一。本文以T/R-R型双基地高频地波雷达系统为背景,重点研究智能化信息融合模型和算法,以及复杂电磁环境下的航迹起始和短航迹合成问题。本文首先针对高频地波雷达电磁波沿海面绕射的特点,采用曲面定位法分析T/R-R型双基地高频地波雷达系统的探测范围及定位精度,绘制单、多测量子集定位精度曲线,并得到双基地高频地波雷达各测量子集在不同探测区域的精度分布。这些研究为双基地高频雷达数据处理方法的设计提供参考依据。受复杂电磁环境、海杂波及电离层杂波等影响,高频地波雷达系统的检测点迹往往具有突变性、非均匀性、区域性和模糊性等特点。本文从信号处理角度,将上述超视距探测时变不确定多源信息融合问题归结为多源、多模式信息的综合决策推理系统设计问题。这类问题的解决需要引入智能化的信息融合概念和算法,从而构造一个具有学习、推理、专家功能的高性能智能化信息融合系统。本文通过分析双基地高频雷达的数据特点及其在数据处理中存在的问题,在神经网络-模糊推理二维模型基础上引入专家系统技术,构造神经网络(Neural Network)、模糊推理(Fuzzy Reasoning)和专家系统(Expert System)三大技术相结合的紧耦合式NFE三维模型。研究表明,该模型能够综合神经网络、模糊推理和专家系统的特点,并具有通用性和收敛性,可看作是一类完备的智能化信息融合模型,适用于处理复杂环境背景下的不确定信息融合问题。由于高频地波雷达的速度探测精度远高于距离和方位的探测精度,因此在关联处理中速度信息起到至关重要的作用。在双基地高频雷达的点迹关联处理中,由于仅凭径向速度无法解决速度对齐问题,进而无法利用目标实际运动速度信息,因此很难预测首次检测点迹下一时刻的位置,这为数据处理的初始相关波门的位置和大小的选取带来极大的困难。针对此问题,本文提出基于NFE模型的初始点迹关联波门预测方法。该方法将回波点迹的径向斜率(由目标位置信息确定)和径向速度作为输入变量,以目标在径向坐标系的预计运动区域相应值作为输出变量,按照双基地高频雷达探测精度的不同建立专家系统和制定专家规则来确定不同区域的隶属函数,实现单检测点迹的初始波门预测,从而为解决双基地高频地波雷达初始点迹的速度关联提出一种可行的方法。该方法为及时准确地确定初始点迹关联波门和快速起始航迹提供有力保障。传统的微波雷达航迹起始算法具有的共性是:其一,主要利用目标的位置信息(包括径向距离和方位角)作为输入向量进行判断,然后考虑速度维的关联处理;其二,这些算法在判别是否形成有效航迹时所遵循的准则多以量测连线的线性程度的判断作为依据。然而双基地高频地波雷达航迹起始问题不能简单地照搬传统的微波雷达航迹起始算法,需要考虑由于低位置信息探测精度引起的检测数据的强突变性问题,从而无法应用传统的方法有效地起始航迹。针对此问题,本文在对双基地高频地波雷达数据特征详尽分析的基础上,研究基于NFE模型的航迹起始方法。该方法建立径向速度变化量、雷达反射截面积变化量及信杂比幅度与回波量测作为更新点迹的可信度之间的模糊关系,间接利用数据的位置信息将其作为专家系统的判决条件作为辅助选择相应的输入变量的隶属度函数实现航迹起始,从而解决双基地高频地波雷达系统在初始点迹高度分散、不连续的情况下航迹难以起始的问题。不同于一般的微波雷达,高频地波雷达采用了超分辨、高精度的多普勒处理技术,因此如何有效利用多普勒频移参数成为双基地高频雷达系统航迹处理的关键。由于在复杂环境下的各站航迹往往是断续的短航迹,而不是完整的稳定航迹,因此双基地高频地波雷达航迹融合实际上是短航迹合成问题。本文针对此问题提出利用多普勒频差因数修正相交关联门内公共量测来自目标的条件概率,并将其引入Hopfield网络的目标函数的Doppler-NJPDA数据关联算法,提高密集杂波环境下交叉运动目标相交的区域正确短航迹识别概率,从而解决双基地高频地波雷达交叉运动目标真假短航迹识别问题。另外,在短航迹合成处理中融汇基于NFE模型的航迹起始方法,并在航迹更新点可信度NFE模型的基础上按照航迹处理的特点对其进行改进用于虚假点迹剔除。综合使用以上方法可保证对多目标点迹-航迹关联的实时性和可靠性,从而解决双基地高频雷达密集虚假点迹环境下交叉运动目标相交区域的短航迹合成问题。更多还原

【Abstract】 High Frequency Surface Wave Radar (HFSWR) works in the short wave band which has crowded frequencies and complicated electromagnetic environments, so its detection performance is susceptible to the ionosphere interference change. The system detection is difficult to guarantee the continuity of the target measurements, and the low azimuth resolution causes the deconcentration and jump of the detected plots. So it is a tough work to initiate target tracks and form stable ones while processing data. However, bistatic/multistatic HF radars are one of the approaches which could increase all-weather performance of monostatic HF radars and ensure the track initiation and forming effective tracks under complicated electromagnetic environments. The track initiation and short track composition under complicated electromagnetic environments is principally researched based on T/R-R bistatic HFSWR system and the intelligentized information fusion models and algorithms is proposed in this paper.According the characteristic that the vertical polarized electromagnetic wave diffracts along the ocean surface, the detection range and localization precision for T/R-R bistatic HF radar system are deduced and simulated using the curved face positioning analysis method, including drawing localization precision curves using single/multiple measurement subsets and the distribution diagram of high precision measurement subsets is obtained. These analyses make the proposing of data processing method based on bistatic HF radar system have pertinence and provide referenced proofs.Affected by the complicated electromagnetic environment, sea clutter and ionospheric clutter, the plots detected by HFSWR have the characteristics of saltation, nonuniformity, deficiency and regionality. The issues above are summarized to be the fusion problems of the uncertain information for bistatic HFSWR system in this paper from the angle of signal processing. The solving of this sort of problems needs to introduce intelligentized information fusion models and algorithms in order to structure a high-powered intelligentized information fusion system in possession of learning, reasoning and expert function. A tightly-coupled NFE model combined with Neural network, Fuzzy reasoning and Expert system is constructed based on neural network-fuzzy reasoning model by introducing expert system technology. It is indicated that this model can be regarded as a category of complete intelligentized information fusion model possessing commonality and approximation and it is adaptive to process the uncertain information fusion problems under complicated environment.The velocity precision is far higher than that of range and direction, so the velocity information takes a significant part in the association processing. It is unavailable to realize velocity coordinate transformation while processing measurement association in bistastic HFSWR in order that the actual velocity information could not be directly utilized, so this make it intractable to choose the position and size of the initial association gate. Aiming at this problem, the initial association gate forecasting NFE model based on single detected plot is proposed in this paper. In this model, the radial gradient (determined by the target position information) and radial velocity are set to be input variables and the predicted movement region in radial coordinate system is set to be output variable. According to the different detection precision for bistatic HFSWR the member functions subject to diverse regions are determined by establishing expert system to realize initial association gate forecasting in order to solve the difficult problem of forecasting target movement trend only depending on the first radar echo for bistatic HFSWR system. This method have not increased the calculation quantity profiting from the parallel processing of neural network as well as increasing the hypothetical gate numbers, so it ensures to determine the initial association gate and initiate tracks opportunely and exactly.The commonness of traditional track initiation methods is: firstly, they all utilize the position information (including radial range and azimuth angle) to make determination and then consider the velocity association; secondly, the criterion for judging whether forming valid tracks is according to the linearity degree of the link line of measurements. This is quite proper and effective for high precision radar system. However, the inherent characteristic of HFSWR determines low detection precision of target position accuracy and this makes the plot data saltation, nonuniformity, deficiency and regionality so that traditional algorithms could not initiate tracks efficiently, and this problem has stood out and desiderates solutions. Based on the considerations above, an NFE track initiation method is proposed from a brand-new angle and idea after a particular analysis of the data features for bistatic HFSWR. It has built the fuzzy relationship between the radial velocity change, radar cross section change and signal to clutter ratio magnitude and the degree of belief for the target measurement which is regarded to confirm the update in the proposed method and the position data are utilized indirectly to choose the corresponding member functions of input variables as the judgement conditions of the expert system, so as to solve the low track initiation probability problems under the circumstance of intermittent measurements in bistatic HFSWR system.HFSWR adopts super-resolution and high-precision Doppler processing technology other than ordinary microwave radar, therefore, how to get sufficient use of Doppler frequency shift parameter is the key of the track processing for bistatic HFSWR. In respect that tracks from each receiver are usually intermittent and short rather than compete and steady, the track fusion for bistatic HFSWR is actually a short track composition issue. Aiming at this situation, the Doppler-NJPDA data association method utilizing doppler frequency shift difference to revise the conditional probability of the public measurements from the targets is proposed to increase measurement-to-track association probability in the intersection of the cross moving targets in the dense false measurements environment in order to solve the true-or-false measurements identification problem of the cross moving targets for bistatic HFSWR. In addition, the adaptive tracking gate algorithm based on detection precision of bistatic HFSWR is proposed in short track composition processing affiliated the NFE track initiation method, and the modified update degree of belief NFE model is used to eliminate the outliers. Integrating the method above, the real-time operation and reliability measurement-to-track association is ensured so as to solve the short track composition problem of the cross moving targets in the intersection for bistatic HFSWR.更多还原