基于改进粒子群优化LS-SVM的卫星钟差预报研究

针对导航卫星短期钟差预报精度和稳定度不高的问题，提出了一种基于改进粒子群优化（PSO）最小二乘支持向量机（LS-SVM）的卫星钟差预报方法。通过引进自适应改变的惯性权重和学习因子来提高粒子群算法的寻优能力，并将其应用到LS-SVM的参数优化中，避免人为选择参数的盲目性，提高了LS-SVM的泛化能力和预报精度。选取国际GPS服务组织（IGS）产品中四颗典型卫星的钟差数据，分别采用LS-SVM模型、神经网络模型和灰色系统模型进行短期钟差预报，计算结果表明：LS-SVM模型的预报精度优于其它两种模型，为导航卫星短期高精度钟差预报提供了新的思路。

LS-SVM Based on Improved PSO for Prediction of Satellite Clock Error

Aiming at the poor performance of short term prediction of navigation satellite clock error, a method based on the least square support vector machine (LS-SVM) and improved particle swarm optimization (PSO) is proposed for prediction of satellite clock error. Adaptive inertia weigh and learning factor are introduced to improve the ability of PSO to find the best swarm. Then it is used to choose the parameters of LS-SVM, for avoiding the man made blindness and enhancing the efficiency of online forecasting. The four typical GPS satellites clock data of IGS are chosen and respectively used in three models to predict short term clock error. The results show that the accuracy of LS-SVM model is superior to the other models, and the work provides a new way for short term prediction of navigation satellite clock error.