Prediction for ATE state parameters based on improved LS-SVM

An improved least squares support vector machines (LS-SVM) was proposed to improve the sparse and robust performance of LS-SVM in the small samples prediction. The sparse and robust performance could be improved through adding elements of weighted LS-SVM and robust LS-SVM. We introduced a contrast experiment for ATE parameters prediction control through the three methods of neural network, LS-SVM and improved LS-SVM algorithm. Simulation results show that the improved LS-SVM algorithm has good performance in ATE parameters prediction, which succeeds in stability assessment for an aviation ATE.