特征根估计方法在SINS制导工具误差估计中的应用

Application of improved latent regression to SINS guidance instrument error estimation

特征根估计方法具有降低模型复共线性程度和提高参数估计效果的作用,但是在SINS制导工具误差估计中,如何有效地筛选特征根却成为特征根估计的重要问题。文中通过研究特征根与残差平方和之间的关系,提出了以残差平方和与系统阶次比最小为准则(即MRSO准则)的特征根筛选方法。仿真分析表明,改进的特征根估计方法具有很强的工程实践性,从而推动了该方法在实际工程中广泛应用。

The Latent Regression of the linear regression can weaken abnormal state and increase the efficiency in estimating parameters. However, how to choose the characteristic root is an important problem for Latent Regression in SINS guidance instrument error estimation. The relation between the characteristic roots and the sum of squares of residual error（SSE） was researched, and the criterion for choosing the characteristic roots was brought out based on Minimum Ratio of SSE and Order（MRSO）. Simulation results show that the improved Latent Regression is more effective and practical.