基于最小二乘支撑矢量机的陀螺仪漂移预测

Gyroscopic Drift Forecasting on Least Squares Support Vector Machine

研究了支撑矢量机的回归估计算法。针对标准支撑矢量机算法训练速度慢、难以解决大规模问题的局限性,对标准算法的约束条件加以改进,得到一种最小二乘支撑矢量机回归估计算法,该算法大大提高了支撑矢量机的训练速度和解决大规模问题的能力。论文将最小二乘支撑矢量机应用于陀螺仪的漂移预测中,仿真实验结果证明了算法的有效性和可行性,为陀螺仪的实时预测及故障预报提供了可能。

This paper researches the regression estimate algorithm with SVM. Pinpoint the problem that the training speed of the standard algorithm is slow and it is hard to solve large scale problem, To overcome this drawback, a Least Squares Support Vector Machine （LSSVM） is put forward in this paper through improving the constraint condition of standard SVM algorithm. The training speed and the ability to solve large scale problem are much improved through using LSSVM. Finally, the LSSVM is applied to forecasting the gyroscopic drift in this paper. The simulated experiment proved the algorithm is valid and acceptable, and it supplies a way to real-Lime forecast and fault - predict of the gyroscopic.