神经网络在光纤陀螺标度因数温补中的应用  

The Application of Neural Network in Temperature Compensation of FOG Scale Factor

作　　者：陈永奇[1] 张春熹[1] 朱奎宝[1] 

机构地区：[1]北京航空航天大学仪器科学与光电工程学院,北京100083

出　　处：《压电与声光》2007年第5期516-518,共3页Piezoelectrics & Acoustooptics

基　　金：二炮预研基金资助项目(103010303)

摘　　要：阐述了光纤陀螺标度因数受温度影响的机理,光纤长度、光纤环直径、中心波长、集成光学调制器调制系数以及2π复位电压值均影响标度因数的稳定性。在此基础上提出建立温度模型补偿法,对最小二乘法和BP神经网络补偿法进行了分析与比较。该文利用光纤陀螺输出实测数据进行了后仿真,结果表明,神经网络法进行光纤陀螺标度因数温度补偿可使误差明显减小,其效果要优于传统的最小二乘法,这样保证了光纤陀螺全温情况下在系统的使用精度。The scale factor of fiber optic gyro is influenced greatly by temperature. This text carried on the analysis about the reason, including the length of fiber, the diameter of fiber coil, middle wavelength, the modulation coefficient of integrated optical modulator and 2π reset voltage. The temperature model compensation was researched deeply. The traditional least-two-multiply method had some disadvantage on solving this problem. BP neural network compensation method was brought forward to this issue. Through the analysis and comparison between these methods, the simulation results indicated that the error reduced greatly by using BP neural network method, which guaranteed actual gyro resolution in inertia navigation system in full temperature scale.

关 键 词：光纤陀螺 标度因数 BP神经网络 

分 类 号：U666.12[交通运输工程—船舶及航道工程]