Online predicting of line-of-sight angular rate based on LS-SVM method

Firstly, the least squares support vector machine (LS-SVM) algorithm was improved by a time-weighted and last-elimination mechanism. Secondly, the improved LS-SVM algorithm was used for line-of-sight angular rate online prediction when the target was unlocked by the seeker due to the line-of-sight of the seeker which was blocked by the clouds (clouds-crossing) or other interferences. On one hand, if the target was locked by the seeker the improved LS-SVM should be used for online training. On the other hand, if the target was unlocked the decision function (result of the online training) should be used for predicting the line-of-sight angular rate. Lastly, by adding up the miss-distances of the numerical simulations in which the seeker was unlocked in the terminal part of the trajectory, the results demostrated the effectiveness of LS-SVM method to the typical line-of-sight angular rate signal predicting and the application prospects in increasing the capacity of clouds-crossing and anti-interference of the small air to surface tactical missies.