基于混沌时间序列的GPS卫星钟差预测算法

GPS satellite clock bias prediction algorithm based on chaotic time series

针对现有钟差预测模型对卫星钟差的非线性特性难以精确预测的特性,提出一种基于混沌时间序列的卫星钟差预测算法。该算法首先通过对钟差序列进行相空间重构,求得最大李雅普诺夫（Lyapunov）指数证明其混沌特性,然后分别采用零阶加权局域预测法和一阶加权局域预测法对钟差序列进行预测,最后将预测结果与国际GNSS服务（IGS）精密钟差值进行比较,得到算法的预测精度。以采样间隔为30 s,时长约为23 h的全球定位系统（GPS）卫星钟差序列进行预测,结果表明15 min内,IGS真实值和预测值的绝对偏差在1 ns以内,绝对偏差平均值在0.3 ns以内。将该算法应用于卫星钟差的预测中,可以实现对卫星钟差非线性特性的短期精确预测。

The current clock bias prediction model can not accurately predict the nonlinear characteristics of satellite clock bias. To deal with this problem,a satellite clock bias prediction algorithm based on chaotic time series is proposed. Firstly,the algorithm reconstructs the phase space of the clock bias sequence and obtains the maximum Lyapunov index to prove the chaotic characteristics. Then,the clock bias sequence is predicted by the zero-order weighted local method and the first-order weighted local method,respectively.Finally,the prediction results are compared with IGS precision clock bias to get the prediction accuracy of the algorithm. The GPS satellite clock bias sequence is predicted with a sampling interval of 30 seconds and a time interval of 23 hours. The results show that the absolute deviation between the IGS true value and the predicted value is within 1 ns and the absolute average deviation is less than 0. 3 ns within 15 minutes. The algorithm can reach the short-term accurate prediction of the nonlinear characteristics of satellite clock bias.