卫星遮挡交通环境下车辆融合定位策略研究

发布时间：2018-11-16 16:45

【摘要】：全球定位系统(GlobalPositioningSystem,GPS)和基于微机电系统(Micro-Electro-Mechanical System,MEMS)技术的惯性导航系统(Inertial Navigation System,INS)的组合为陆地车辆的准确、可靠定位提供了一种低成本解决方案。但是在GPS中断期间,MEMSINS/GPS组合系统的定位精度迅速降低。为了解决此问题,本文开展了卫星遮挡交通环境下车辆融合定位策略研究,围绕MEMS惯性测量单元(Inertial Measurement Unit,IMU)随机误差处理和 MEMSINS/GPS信息融合算法两大关键技术,针对其中的若干关键性问题展开深入探索。主要研究内容及成果包括:(1)为了抑制MEMS IMU随机误差的高频部分,本文设计了一种基于改进小波滤波(Improved Wavelet Filter,IWF)的数据预处理算法。针对传统小波滤波中分解层数确定以及阈值函数存在的不足,该算法采用频谱分析结合实验评估的方法来选择小波分解的最优层数,并设计了一种改进的带调节因子的阈值函数。实验结果表明,该方法在一定程度上克服了传统阈值函数的缺点,可以较好的抑制MEMS惯性数据随机误差的高频部分。(2)针对MEMS IMU随机误差的低频缓变部分,本文结合经验模态分解(Empirical Mode Decomposition,EMD)和分形高斯噪声(fractional Gaussian noise,fGn)的优点,设计了一种基于EMD区间阈值滤波的自适应数据预处理算法。该算法首先利用fGn对惯性数据中的随机误差进行建模,然后利用fGn在各层本征模函数(Intrinsic Mode Function,IMF)内方差的关系,确定了 EMD滤波阈值选择准则;同时,构建了一种基于IMF区间的阈值处理方案,来消除滤波后信号的不连续性。实验结果表明,该预处理算法既可以有效的消除部分低频缓变随机误差,又可以消除随机误差的高频部分,与基于改进小波滤波的预处理算法相比,滤波后惯性数据数据的精度得到进一步提升。(3)提出了一种基于自回归(Auto-Regressive,AR)模型辅助卡尔曼滤波(KalmanFilter,KF)的混合策略,来提高卫星遮挡交通情况下车辆组合定位系统的精度和可靠性。该策略首先在系统结构中引入了惯性数据预处理步骤,为后续信息融合提供高精度的惯性数据;然后对传统的INS误差建模结构进行改进,设计了一种基于序列的INS误差建模和预测结构;在此基础上,设计了基于AR模型的前向估计器(AR model-based Forward Estimator,ARFE),并构建基于ARFE/KF的混合策略对INS位置误差建模并预测。实车实验结果表明,该方法可以有效地补偿GPS中断情况下INS位置误差,并具有较好的泛化能力和实时性,大幅度提高了 GPS遮挡情况下的车辆定位精度。(4)为了进一步提高车辆组合定位系统在较长GPS中断情况下的性能,提出一种基于最小二乘向量机(Least Squares Support Vector Machine,LSSVM)的带外部输入非线性自回归模型(Nonlinear Auto-Regressive with Exogenous inputs,NARX)辅助KF的INS误差混合预测策略。该策略设计了一种带有记忆功能和内部反馈的INS误差建模结构,兼顾了 INS误差历史发展趋势和车辆运动状态影响;构建了 LSSVM-NARX/KF混合策略对INS误差进行建模,并在GPS中断期间实现对INS误差的预测和补偿。实车实验表明,该方法对各种驾驶工况具有较好的适应性,可以有效的抑制INS定位误差的积累,能在GPS发生较长时间中断的情况下为车辆提供更加准确、可靠的定位信息。
[Abstract]:The combination of Global Positioning System (GPS) and inertial navigation system (INS) based on Micro-Electro-Mechanical System (MEMS) technology provides a low-cost solution for the accurate and reliable positioning of land vehicles. However, the positioning accuracy of the MEMSINS/ GPS combined system is rapidly reduced during the GPS interruption. In order to solve this problem, this paper has carried out the research of the vehicle fusion positioning strategy in the environment of the satellite-shielded traffic environment, the two key technologies of the random error processing of the MEMS inertial measurement unit (IMU) and the MEMSINS/ GPS information fusion algorithm, and the deep exploration of some of the key problems. The main research contents and achievements include: (1) In order to suppress the high-frequency part of the random error of the MEMS IMU, a data preprocessing algorithm based on improved wavelet filter (IWF) is designed. In the light of the limitation of the determination of the number of decomposition levels and the existence of the threshold function in the traditional wavelet filtering, the optimal number of layers of the wavelet decomposition is selected by the method of spectral analysis and experimental evaluation, and a modified threshold function with an adjustment factor is designed. The experimental results show that the method overcomes the shortcomings of the traditional threshold function to a certain extent, and can better restrain the high-frequency part of the random error of the MEMS inertial data. (2) Based on the advantages of empirical mode decomposition (EMD) and fractal Gaussian noise (fGn), an adaptive data preprocessing algorithm based on EMD interval threshold filtering is designed. The method firstly uses fGn to model the random error in the inertial data, then uses fGn to determine the relationship between the variance of the intrinsic mode function (IMF) of each layer, and determines the EMD filtering threshold selection criterion; and meanwhile, a threshold processing scheme based on the IMF interval is constructed, to eliminate the discontinuity of the filtered signal. The experimental results show that the pre-processing algorithm can effectively eliminate the random errors of some low-frequency and random errors, but also can eliminate the high-frequency part of the random error. Compared with the pre-processing algorithm based on the modified small-wave filtering, the accuracy of the post-filter inertial data data is further improved. (3) A hybrid strategy based on the self-regressive (AR) model-assisted Kalman filter (KF) is proposed to improve the accuracy and reliability of the vehicle-combined positioning system in the case of the satellite blocking traffic. The method comprises the following steps of: firstly, introducing an inertia data preprocessing step in a system structure to provide high-precision inertial data for subsequent information fusion; then, improving the traditional INS error modeling structure, and designing a sequence-based INS error modeling and prediction structure; and on the basis, An AR model-based forward estimator (ARFE) is designed and a mixed strategy based on ARFE/ KF is designed to model and predict the position error of the INS. The results of real-vehicle experiment show that the method can effectively compensate the position error of the INS in the case of GPS interruption, and has better generalization ability and real-time performance, and greatly improves the positioning accuracy of the vehicle under the condition of GPS shielding. (4) In order to further improve the performance of the vehicle-combined positioning system under the condition of longer GPS interruption, an INS-error mixed prediction strategy based on an external input non-linear auto-regressive model (NARX)-assisted KF based on the Least Squares Support Vector Machine (LSSVM) is proposed. In this paper, an INS error modeling structure with memory function and internal feedback is designed, and the development trend of INS error and the influence of vehicle motion state are taken into account; the INS error is modeled by the LSSVM-NARX/ KF mixing strategy, and the prediction and compensation of the INS error is realized during the GPS interruption. The real-vehicle experiment shows that the method has better adaptability to various driving conditions, can effectively inhibit the accumulation of the INS positioning error, and can provide more accurate and reliable positioning information for the vehicle under the condition that the GPS is interrupted for a long time.
【学位授予单位】：东南大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：U463.67;U495

【相似文献】