摘要
激光陀螺随机漂移数据易受环境的影响,特别是温度因数,其输出表现为时间与温度的非线性关系。传统的对数据建模方法包括时间序列和神经网络方法,但单纯利用这两种方法对诸如随机漂移这种波动性较大的数据建模精度不够高。运用灰色理论对原始信号进行预处理,得到规律性较强的累加数据;再利用时序和神经网络法进行建模,提出了两种组合建模方法:灰色时序(GARMA)建模法和灰色神经网络(GRBFN)建模法,并将其运用到漂移数据的处理中。仿真结果表明,提出的组合模型拟合精度高于任何一种单独建模效果。
The laser gyroscope random drift is liable under the influence of environment,especial the temperature and its outputs can be expressed with non-linear membership of time and temperature. The traditional methods of modeling embraced the time serial and neural network,but the degree of accuracy of using merely these two methods to model the fluctuation data such as random drift is not very high. The grey theory was used to preprocess the original signal to generate accumulated data with better regularity,and then time serial and neural network were utilized to model. Two combined modeling methods of GARMA and GRBFN were presented and were used to process the drift data. The simulation results showed that this new combined model approach had higher precision than any other single patterns.
出处
《压电与声光》
CSCD
北大核心
2009年第6期800-803,共4页
Piezoelectrics & Acoustooptics
基金
国家自然科学基金资助项目(60304004)
