基于改进PSO优化LSSVM的MEMS陀螺随机漂移预测  

Forecasting of MEMS Gyroscope's Random Drifts Based on LSSVM Optimized by Modified PSO

作　　者：孙田川 刘洁瑜[1] 康莉 杨浩天 

机构地区：[1]第二炮兵工程大学控制工程系,西安710025 [2]二炮驻699厂军代室,北京100039

出　　处：《传感技术学报》2016年第6期854-858,共5页Chinese Journal of Sensors and Actuators

摘　　要：提出一种基于改进粒子群算法(PSO)优化最小二乘支持向量机(LSSVM)的MEMS陀螺随机漂移的预测模型建立方法。该方法首先应用最小二乘支持向量机对MEMS陀螺随机漂移建立预测模型,然后应用改进粒子群算法对该模型进行优化,最后应用参数优化后的LSSVM预测模型对随机漂移进行预测。该方法不仅解决了支持向量机训练速度慢和所需计算资源多的问题,而且文中提出的改进的惯性权值递减策略使PSO算法在全局或局部搜索能力上的侧重具有更好的适应度。实验结果表明,该预测模型可以有效地进行陀螺随机漂移的预测,且预测效果优于基本PSO优化的最小二乘支持向量机。The predictive-modeling method for the random drift of MEMS gyroscope is proposed,which is based onthe least squares support vector machine optimized by modified particle swarm algorithm. Firstly,a forecasting model of the random drift of MEMS gyroscope is built with least squares support vector machine,and then,the modifiedparticle swarm algorithm is put forward to optimize the model. Finally,the optimized LSSVM model is used to pre-dict the random drift of MEMS gyroscope. The modeling method solves SVM＇s disadvantage of slow training speedand requesting more resources. What＇s more,the modified PSO have better fitness in capability of global searchingor local searching. The experimental result also shows that the new modeling method can effectively predict MEMSgyroscope＇s random drifts,and it has better effect than LSSVM optimized by PSO.

关 键 词：微机械陀螺仪 最小二乘支持向量机 改进粒子群优化算法 随机漂移 

分 类 号：V241.5[航空宇航科学与技术—飞行器设计]