姿态角高精度数据预测方法

High-precision Data Prediction Methods of Attitude Angle

针对机载时空联合调制傅里叶变换成像光谱仪在对地观测过程中数据质量易受运动误差影响,而定位定姿系统和惯性稳定平台组合中定位定姿系统与成像光谱仪的时间存在不同步的问题,提出一种基于最小二乘支持向量机插值预测得到同步姿态数据的方法。算法通过非线性变换将输入向量映射到高维特征空间,利用最小二乘函数和mercer条件得到优化回归模型,并构建二维网格平面以确定模型参数最优解。仿真结果表明,预测得到的姿态数据平均绝对误差最大为0.001 1°,均方根相对误差最大为1.48%,与以往采用的插值方法相比,现有算法精度有显著提高。

To solve the problem that the quality of data from airborne Temporally and Spatially Modulated Fourier Transform Imaging Spectrometer was easily affected by motion error during the process of earth observation,and there was mismatch between the measure data of the Position and Orientation System and the sampling frame frequency of the imaging spectrometer by using Position and Orientation System and Inertial Stabilization Platform, an interpolation frame frequency matching method based on least squares support vector machines was proposed. The input vectors were mapped into the high- dimension feature space by non linear transformation, and the optimal solution of model parameter was obtained through two- dimension mesh plane constructed by optimized regression model, which was set up by least squares function and mercer condition. The simulation result indicates the maximum average absolute error of the interpolated frame frequency matching data is 0.11% ,the maximum relative error of the root mean square is 1.48% ,which makes notable improvement compared with the interpolation method used in the previous literature.