【作者】 徐田来；

【导师】 崔平远；

【作者基本信息】 哈尔滨工业大学 ， 飞行器设计， 2007， 博士

【摘要】 导弹信息化机动发射平台的应用环境非常复杂,为了能够实现导弹的快速机动发射,不仅要求导航系统能够提供高精度的导航信息,还要求导航系统具有高度的可靠性、自主性和抗干扰性。任何单一导航系统或简单组合的导航系统都难以满足这一要求,因而,多传感器信息融合技术成为导航系统研究的主要技术途径。本文以导弹信息化机动发射平台为应用背景,以提高导弹信息化机动发射平台导航系统定位精度和在复杂环境中的适应能力为目标,研究了基于信息融合的多传感器组合导航技术与GPS(全球定位系统)失效时的性能辅助技术,有针对性地对基于信息融合技术的多传感器组合导航系统进行设计和开发。本文主要研究内容包括:研究了INS(惯性导航系统)/GPS组合导航自适应滤波算法。针对车载INS/GPS组合导航量测噪声统计特性可能随应用环境不同而发生变化的问题,从基于新息的角度和基于多模型的角度着手,研究了两种自适应滤波算法。一种算法为模糊自适应卡尔曼滤波算法,该算法根据实时得到的量测新息的实际方差与理论方差的比值,由设计的模糊推理系统在线调整模型中的量测噪声协方差矩阵。另一种算法为自适应交互多模型算法,该算法将改进的Sage-Husa自适应滤波器与交互多模型相结合,以Sage-Husa自适应滤波的估值为中心,扩展得到交互多模型算法的模型集,以少量的模型实现对实际模态的覆盖。研究了置信度加权的多传感器模糊自适应数据融合算法。首先在联合卡尔曼滤波算法中使用模糊自适应滤波代替标准卡尔曼滤波,子滤波器自适应的跟踪实际量测噪声统计特性,构成模糊自适应联合卡尔曼信息融合算法。然后在该算法的基础上,利用置信度对各子滤波器及联合卡尔曼滤波器进行加权,实现置信度加权的模糊自适应数据融合算法。为了提高INS/GPS系统在GPS无效时的定位精度,研究了将车辆运动学约束和路网约束作为虚拟传感器的测量,辅助INS/GPS的方法。车辆正常行驶时,理想情况下侧滑速度为零,此速度约束是车辆行驶时固有的运动学约束,被用来构造一个虚拟传感器的测量;另一方面当车辆在路网上行驶时,在地图匹配方法基础上,由路网约束构造另一个虚拟传感器的测量。当GPS无效时,在满足应用条件的情况下,使用虚拟传感器辅助INS,以提高INS/GPS系统在GPS无效时的定位精度。进行了基于Matlab/RTW/xPC的车载组合导航半实物仿真平台的设计,使用一台Matlab/Simulink/RTW宿主机和两台xPC目标机的系统结构,实现组合导航的实时仿真。结合以上多传感器信息融合算法和虚拟传感器辅助的INS/GPS组合导航技术,进行了多传感器组合导航系统的整体设计和软、硬件的研制。并进行了多传感器组合导航系统的跑车实验,实验结果证明了本文所研究算法的有效性和所设计的多传感器组合导航系统的可用性。更多还原

【Abstract】 In order to launch missile rapidly and movably under complex application environment, Information-based missile-launcher requests the navigation system not only to be accurate in providing the navigation information, but also to be high reliable, independent and anti-jamming. It is difficult to meet this requirement for any single or the simple combination navigation system. Therefore, the multi-sensors data fusion technology has provided the technical support for building a navigation platform with high quality. Using Information-based missile-launcher as the application background, aiming at improving Information-based missile-launcher positioning accuracy as well as the adaptive and the fault-tolerant ability in the complex environment, thorough research on the information fusion algorithm and INS/GPS performance enhancement technology based on multi-sensors was conducted. The major contents were summed up as follows:INS/GPS measurement noise statistical characteristics may differ with the application of environmental change, to deal with this problem, on the basis of Sage-Husa adaptive filtering and interactive multi-model estimation theory, two adaptive filtering algorithms are presented. One is fuzzy adaptive Kalman filtering algorithm. By monitoring the ratio between filter residual and actual residual, this algorithm modifies recursively the measurement noise covariance of Kalman Filtering online using the Fuzzy Inference System (FIS) to make the covariance close to real measurement covariance gradually. The other is adaptive interactive multiple model (AIMM) algorimth. AIMM combined IMM algorithm with the improved Sage-Husa adaptive filtering algorithm. AIMM algorithm can achieve the coverage of real situation through few sub-models, and the accuracy can be improved than IMM algorithm.The fuzzy adaptive Kalman information fusion algorithm based on confidence is presented. Firstly, fuzzy adaptive filter instead of the standard Kalman filter is used as sub-filter of federated Kalman filter, form fuzzy adaptive federated Kalman information fusion algorithm. The sub-filters can adaptive track the actual measurement noise statistical characteristics. Secondly, fuzzy adaptive Kalman information fusion algorithm is combined with method of confidence weighted. The algorithm on the one hand makes measurement noise covariance of sub-filters adaptive track actual measurements of noise statistics, on the other hand can automatically lower the confidence of low filter weights.To improve the accuracy of INS/GPS when GPS outages occur, kinematic constraints of land vehicle and road network constraints are regarded as two virtual sensors, used to enhance the performance of INS/GPS during GPS outages. Under ideal conditions, velocity of the vehicle in the plane perpendicular to the forward direction is zero, means there is no side slip. The constraints can be used to form measurements of a virtual sensor. Constraints of road network can be used to form measurements of the other virtual sensor based on map matching when the vehicle is on road networks. When GPS is unavailable, the two virtual sensors is used to aid INS, the accuracy of INS/GPS can be improved.Based on Matlab/RTW/xPC the Vehicle Integrated Navigation HWIL simulation platform is designed, the structure of one Matlab/Simulink/RTW host and two xPC target machine is adopted, and the real-time simulation of integrated navigation is achieved.Based on multi-sensor information fusion algorithm and the INS/GPS performance enhancement technology by virtual sensors, software and hardware of multi-sensor integrated navigation system are developed. The experimental results proved that the algorithms presented in this dissertation are valid and the multi-sensor integrated navigation system designed in this dissertation is practicable.更多还原