基于改进RBF神经网络的激光陀螺温度补偿方法  

A Method for Compensating Temperature of LG Based on Improved RBF Neural Networks

作　　者：赵琳[1] 张健[1] 陈帅[1] 丁继成[1] 

机构地区：[1]哈尔滨工程大学自动化学院,哈尔滨150001

出　　处：《弹箭与制导学报》2014年第6期17-20,24,共5页Journal of Projectiles,Rockets,Missiles and Guidance

基　　金：国家自然科学基金(61273081;61304234)资助

摘　　要：针对温度及温度变化对激光陀螺仪温度漂移的影响,在RBF神经网络的基础上,文中提出了基于Kohonen神经网络和正交最小二乘(OLS)算法的RBF神经网络温度补偿方法。该方法可以快速、准确辨识激光陀螺的温度漂移。文中分别采用改进RBF神经网络法、传统RBF神经网络法及多元线性回归法,分别进行辨识与补偿试验。结果表明,经过改进RBF神经网络模型补偿后的陀螺零偏能够满足陀螺恒温测试精度要求,而且避免了陀螺输出受温度及温变速率的影响。To compensate temperature error of laser gyroscope,an improved RBF neural network based on Kohonen neural networks and OLS algorithm was presented. The temperature error can be identified faster and more accurately. The identification and compensation tests have been conducted through the improved RBF neural networks method,the traditional RBF neural networks method and multiple linear regression method. The results show that the compensated bias of laser gyroscope can achieve precision test in the constant temperature and avoid the impact of gyro temperature and temperature change.

关 键 词：KOHONEN神经网络 正交最小二乘 激光陀螺仪 温度补偿 

分 类 号：V241.5[航空宇航科学与技术—飞行器设计] V249.322