支持向量回归辅助的GPS/INS组合导航抗差自适应算法

卡尔曼滤波残差分量受到观测信息误差和动力学模型误差的双重影响，由于GPS/INS松耦合导航系统中观测值个数少于状态参数个数，导致异常检测时难以正确区分误差来源，提出一种支持向量回归辅助的组合导航抗差自适应算法。该算法克服了组合系统观测信息无冗余情况下异常检测的局限性，基于遗传算法参数寻优构建回归模型，预测次优观测值，结合整体异常检验法自主选择抗差或自适应滤波，进而调整观测值或动力学模型对导航解的贡献，进行导航预报。最后利用车载实测数据进行验证，结果表明：该算法能够对存在的异常故障智能判定，减弱观测值异常和动力学模型误差影响，保证组合导航精度，提高导航解可靠性。

SVR Aided Adaptive Robust Filtering Algorithm for GPS/INS Integrated Navigation

The number of observations is less than the number of state parameters in loosely-coupled global positioning system and inertial navigation system (GPS/INS) integrated navigation system. It is hard to distinguish dynamical model error from observation gross error using observation and state residuals, resulting from that the residuals are affected by both dynamical model error and observation gross error. A robust adaptive kalman filtering (RAKF) algorithm is put forward based on genetic algorithm and support vector regression (GA-SVR). The algorithm addresses the limits of anomaly detection on condition of lacking redundant observations. Support vector regression algorithm is used to train the mapping model for predicting suboptimal observations with parameter optimization based on genetic algorithm. The global abnormal detection, combined with the predicted observations, choose robust or adaptive kalman filtering autonomously for purpose of adjusting contribution of observations and dynamical model to the results. Finally field data on the vehicle are collected to verify the algorithm. It's shown that, dynamical model error can be distinguished from observation gross error based on GA-SVR, the influence of anomaly data is greatly weakened with RAKF algorithm to improve the reliability and accuracy of navigation solutions.