轴向陀螺漂移是影响单轴旋转惯导系统导航精度的主要因素。对于轴向陀螺漂移的预测, 提出了一种基于支持向量机的算法。利用初始对准12 h内系统纬度误差和温度变化量作为训练数据, 构造了以多项式、径向基、小波函数为核函数的支持向量机、最小二乘支持向量机、遗忘因子最小二乘支持向量机, 对比了它们用于轴向陀螺漂移预测的泛化性能。试验结果表明: 遗忘因子最小二乘支持向量机可有效地用于轴向陀螺漂移预测, 具有很高的预测精度, 极大地提高了单轴旋转激光陀螺惯导系统的导航精度。

The axial drift of ring laser gyroscopes (RLGs) is a major factor that impacts the performance of the single-axis rotation inertial navigation system. A square support vector machine (SVM) algorithm to forecast the axial RLG drift is proposed in the paper. Latitude and temperature variation during the identification stage are adopted as inputs of model. Employing polynomial, radial basis function (RBF) and wavelet function as the kernel functions, we construct three kinds of support vector machine (SVM), which are SVM, least square support vector machine (LSSVM) and forgetting factor least square support vector machine (FFLSSVM) and compare their capability to forecast the axial RLG drift. The navigational results show that the FFLSSVM method is an effective approach and has the best identification precision. The experimental result of navigation can meet the practical demand.