【作者】 曹梦龙；

【导师】 崔平远；

【作者基本信息】 哈尔滨工业大学 ， 飞行器设计， 2009， 博士

【摘要】 车载导航系统在理论研究与工程应用上向着智能化与可视化的方向发展。常用的导航方案一般都采用GPS/INS的组合方式,但其理论方法与应用研究工作几乎都集中在GPS和INS都可用条件下对卡尔曼滤波算法的替代或改进方面。在GPS信号无法覆盖以及INS精度下降情况下,如何采用低成本的系统重构方案以及智能信息融合算法来提高导航系统的可靠性与保证系统的导航精度是需要尽快解决的研究课题。论文结合国家“863”计划项目“组合导航系统自主重构技术与智能导航算法研究”对车载导航系统重构技术与数据融合算法问题进行深入的研究。主要研究内容包括以下几个方面:研究了车载导航系统的故障检测与隔离问题。在分析车载导航系统故障检测问题的基础上,采用小波技术对各传感器的状态信号作小波分析,在较短时间内发现故障点。通过模糊推理的方法在故障系统和非故障系统之间进行无扰动切换,隔离存在故障的系统,实现自适应系统重构。研究了室外环境中基于点特征的同时定位与地图构建(Simultaneous Localization and Mapping-SLAM)算法。针对SLAM算法中存在的计算复杂度与信息丰富度之间的矛盾,提出提取环境陆标点特征并转化为线特征的P-L(Point to Line)地图构建算法,形成SLAM地图构建中室外环境信息表达的新方法。为提高地图创建方法的适应能力,从降低观测信息的不确定性入手,利用道路约束条件对SLAM算法中的状态向量进行估计,减少定位误差,增强地图创建的精确度。研究了GPS信号失效时采用低成本组合方式进行系统重构的问题,设计基于地图构建的系统重构方案、基于协作伪卫星的系统重构方案,并完成可应用性分析。研究了基于多模型的噪声自适应技术和非线性噪声耦合技术,用以提高系统状态估计的精度。将人工智能技术融合到尔曼滤波法中去,设计基于神经网络的自适应因子,用以调节滤波器增益,提高滤波稳定性。在上述研究成果的基础上,以机动车沿特定的线路运动过程为基本考察对象,在车辆运动过程中进行卫星导航信号和惯性导航信号的采集、同步和处理,对提出的方案及算法实现动态过程模拟,验证方案及算法的有效性。更多还原

【Abstract】 Vehicular navigation is launched on its ways to intelligentization and visualization. The terms of GPS/INS integrated navigation is generally used on the conventional navigation plan whose theoretical approaches and applicational researches focus more on replacement or improvement of Kalman filtering algorithm only if the GPS and INS are both available. Subject to the coverage of the signal of GPS and the precision of the INS, however, how to improve the reliability and precision of the navigation system using low-cost system reconstruction scheme and intelligent information fusion algorithm is a research subject which needs urgent solution.With the supports of the 863 Program‘Study on system reconstruction and intelligent information fusion algorithm of integrated navigation’, this dissertation deeply studies the technique of vehicular navigation system reconstruction and the algorithm of information fusion. The main contents of this dissertation are as follows.Firstly, the technology of the fault detection and isolation of vehicular navigation is studied. The state signals of the sensors are conducted with the wavelet analysis in order to detect the malfunction in a short time, which is based on analyzing the problem of fault detection for the vehicular navigation system. Non-disturbance switch from the failure system to the non-fault system is executed using the fuzzy reasoning method, which can isolate the fault point, and then, the adaptive reconstruction of system is realized.Secondly, the point to line P-L Simultaneous Localization and Mapping (SLAM) algorithm which is based on the outdoor environment is researched. For the conflict between the computational complexity and the information richness mapping Algorithm which transforms the point features of environmental signs into the line characteristics is proposed to form a new method to express the information of outdoor environment. As starting with reducing the uncertainties of observable information, the state vectors of SLAM algorithm are estimated using the constraints of roads, which can reduce the positioning error and increase the map accuracy, in order to enhance the adaptive ability of map creation.Next, the issue of system reconstruction using low-cost compound mode when the signal of GPS is unavailable is worked. The system reconstruction programs based on the map building and synergic pseudo-satellite are designed and the feasibility is analyzed.Then, the technologies of noise adaptive based on the multi-model and nonlinear noise coupling are investigated to increase the accuracy of estimation of system states. Neural network-based adaptive factors are designed, which incorporate the Kalman filtering algorithm with the artificial intelligence technology, to adjust filter gain and improve the filter stability.Lastly, on the basis of the above research results, focusing on the procedure of the motor vehicle moving along a specific trajectory, the dynamic simulation is conducted, with acquiring, synchronizing and manipulating the signals of the satellite navigation and inertial navigation acquisition during vehicle moving, to illustrate the efficaciousness of the programs and algorithms which is proposed.更多还原