论文部分内容阅读
角跟踪误差是一典型的非平稳随机过程。传统的分析是分段平稳化处理,但这种方法分析的精度较低。本文建立了一种时变AR数学模型,通过把模型系数用距离的拉格朗日多项式基展开,从而把模型系数的时变性转化为基的时变性。论文还讨论了正交辨识和最小二乘在线递推两种辨识算法,并对后者的实时性作了较详细的分析。利用时变 A R数学模型,对一仿真角跟踪误差序列进行了建模,结果同理论值非常吻合。论文还对一实测数据进行了建模和分析,并把结果同传统分段平稳化分析结果作了比较。
Angle tracking error is a typical non-stationary random process. The traditional analysis is piecewise smoothing, but the accuracy of this method is low. In this paper, we establish a time-varying AR mathematical model, which transforms the time-varying of the model coefficients to the time-dependent of the base by expanding the model coefficients by the distance Lagrange polynomials. The dissertation also discussed two recognition algorithms of orthogonality identification and least square recursive recursion, and made a detailed analysis of the latter’s real-time performance. Using time-varying A R mathematical model, a simulation of angular tracking error sequence was modeled, the result is in good agreement with the theoretical value. The paper also made a modeling and analysis of the measured data, and compared the results with the results of the traditional piecewise stability analysis.