基于粒子群优化算法的LS-SVM的GPS高程拟合

GPS height fitting based on PSO optimizing LS-SVM

在GPS高程拟合中,传统拟合方法存在多数据、过学习、泛化能力弱等缺点,导致拟合结果精度欠缺,为此提出了LS-SVM拟合模型。利用粒子群算法对LS-SVM模型的初始参数进行了优化,通过实测数据对该模型进行了分析。实验结果表明,基于粒子群算法优化的LS-SVM模型较传统单一的二次曲面拟合法、BP神经网络、LS-SVM等模型拟合精度高。

In traditional fitting methods there are many shortcomings,such as＂multi data,over learning,weak generalization ability＂,with less accuracy of fitting results. A method is presented to establish a least squares support vector machine model based on particle swarm optimization algorithm. Through the experimental data accuracy of the analysis to the model,compared with the traditional single model of quadric surface fitting method,the BP neural network,the LS-SVM,the experimental results show that the LS-SVM model has a higher precision based on particle swarm optimization algorithm.