中文摘要：

设计弹道目标的实时跟踪滤波器时,精确的动态建模和测量系统误差自校准是提高滤波精度的重要手段。基于补偿最小二乘原理和Tikhonov正则化定理,将难于参数化的复杂运动成分和测量系统误差表示为非参数分量,建立了基于半参数化建模的弹道递推模型和实时滤波算法,并详细设计了半参数建模时所需的光滑因子、正则化矩阵和窗口宽度等建模参数的在线选取方法。仿真实验和实际数据的计算结果表明,非参数成分可以有效地补偿状态模型误差,且可在线校准测量系统误差,显著改善弹道估计精度。算法已初步应用于实际飞行任务,其中的递推建模方法和参数选择策略对其它实时滤波器的设计同样有参考价值。

英文摘要：

Accurate dynamic modeling and self-calibration of systematic error are efficient techniques for designing a real-time filter in trajectory target real-time tracking tasks.Based on the compensative least square method and the Tikhonov regularization theorem,non-parametric portions are added to express the complicated dynamics characteristics and the systematic measurement errors.According to these expressions,a real-time filter for trajectory target tracking is developed on the basis of the recursive dynamic model.Then,the adaptive techniques for selection of the smooth factor,the regular matrix and the length of the sliding window are discussed in detail.The approaches here can notably improve the estimate accuracy and the self-calibration ability of systematic error for trajectory tracking,and are validated by both simulation and real flight data.Actually,the filter has already been applied in some real flight task.Moreover,approaches for recursive dynamic modeling and parameter selection can also benefit the design of other real-time filters.