基于LS-WSVM的光纤陀螺漂移辨识

Identification drift of FOG based on least-square wavelets support vector machine

提出了一种最小二乘小波支持向量机(LS-WSVM)的光纤陀螺的漂移辨识算法。该方法将Mexihat小波函数作为核函数,与最小二乘支持向量机(LS_SVM)相结合建立起通用模型;用光纤陀螺漂移数据训练通用模型,从而得到该光纤陀螺的漂移模型。并用F法则检验了该模型的适应性。试验表明,在相同条件下,与基于Gauss核函数的最小二乘支持向量机模型相比,该模型拥有更高的辨识精度。证明了用最小二乘小波支持向量机对光纤陀螺的机漂移辨识是合适的,有效的。

A new method of modeling FOG drift data using the Least Squares Wavelets Support Vector Machine （LS-WSVM） was proposed. The general model was established by combining the Mexihat wavelet function, which is the kernel function, with the Least Squares Support Vector Machine （LS-SVM）. By training the general model with the drift data, the FOG drift model was obtained. The model was verified by the F rules. The results indicate that this model gives higher identification precision than LS-SV When the Gauss kernel function is in the same condition, which validates the effectiveness of the LS-WSVM method.