基于改进的粒子群优化的北斗卫星钟差组合预报方法

An improved particle swarm optimization method for predicting ultra-fast clock errors of BeiDou satellites

针对现有卫星钟差预测模型预报精度不高的问题,本文构建一种将集合经验模态分解和改进的粒子群优化的随机森林预测算法来预测卫星钟差数据,并采用GPS、北斗钟差数据验证算法的正确性。利用改进的粒子群优化的随机森林模型对集合经验模态分解后的各分量进行预测,并将各分量预测的结果叠加为最终预测结果。最终通过北斗钟差数据取得优良的预测结果,与ISU-P的数据比较,精度提升幅度在12.7%~21.5%之间。

In view of the low accuracy of the existing satellite clock error prediction model,this paper constructs a random forest prediction algorithm based on the set empirical mode decomposition and improved particle swarm optimization to predict the clock errors of BeiDou satellite,and uses the GPS clock error to verify the accuracy of the algorithm.The improved method is used to predict the components after the set empirical mode decomposition,and the prediction results of each component are superimposed into the final prediction results.Finally,good prediction results were obtained through the BeiDou clock difference data.Compared with the ISU-P data,the accuracy was improved by 12.7%~21.5%.