基于最优LS-SVM的制导工具误差分离与折合

将最小二乘支持向量机方法应用于制导工具误差分离于折合。利用线性核函数获得了工具误差系数的估计,然后利用交叉验证技术推导了最小二乘支持向量机最优参数的选择准则。该准则的计算是基于模型求解的中间参数,所以并没有增加很多的计算量。最后根据六自由度弹道仿真软件进行了特殊弹道与全程弹道的仿真。仿真计算表明,与最小二乘和主成份方法相比,最优最小二乘支持向量机获得的误差系数估计与真值更加接近,折合得到的全程弹道遥外差更加准确。

Separation and conversion for guidance instrumentation error using optimal least squares support vector machines

The least squares support vector machines(LS-SVM) are applied to separating and converting the guidance instrumentation systematic error(GISE),and the estimation of GISE is obtained based on the linear kernel function.Then,a simple and efficient criterion is described to model selection based on leave-one-out cross-validation for leastsquares support vector machines.Since the calculation of the criterion only depends on the by-product of the training algorithm,the computational expense of the leave-one-out cross-validation procedure becomes increasingly negligible as the training set becomes larger.The six-freedom trajectory simulation results show that the optimal LS-SVM method has a higher precision than the least square method and principal component method.