基于光纤传感航天器神经网络与数字孪生研究
Research on Spacecraft Neural Network and Digital Twin Based on Optical Fiber Sensing
范丽 1胡泽阳 1武丹 2梁纪秋 2胡夏芬 2张芸山 3谢久富1
作者信息
- 1. 浙江大学湖州研究院
- 2. 湖北航天技术研究院总体设计所
- 3. 北京工业大学物理与光电工程学院
- 折叠
摘要
文中设计了一种掺杂Al2O3、Y2O3、P2O5的新型石英光纤,制作的光纤成品纤芯折射率类线性分布.通过飞秒激光直写技术,在光纤上刻写出光纤Bragg光栅,制作出光栅串.将多根光栅串布设在航天器卫星模型上,形成航天器智能蒙皮,用于感知卫星结构损伤状态.基于上述测量数据,使用unity3D软件开发航天器的3D模型及相匹配的python程序对数据进行处理,实现航天器的数字孪生和状态感知.进一步通过模型建立和BP神经网络算法对传感数据进行感知训练,模型对撞击信号的预测准确率高达90％.
Abstract
A new type of silica fiber doped with Al2O3,Y2O3 and P2O5 was designed,and the refractive index distribution of the fiber core was linear.The fiber Bragg grating was written on the fiber and the grating string was made by femtosecond laser di-rect writing technique.A plurality of grating strings was arranged on a spacecraft satellite model to form a spacecraft intelligent skin which was used for sensing the damage state of a satellite structure.Based on the above measurement data,unity3D software was used to develop a 3D model of the spacecraft and a matching python program to process the data,so as to realize the digital twinning and state perception of the spacecraft.The model was built and BP neural network algorithm was used to train the sens-ing data.The accuracy of the model to predict the impact signal is up to 90％.
关键词
光纤传感/数字孪生/飞秒激光光刻/智能感知/BP神经网络Key words
optical fiber sensing/digital twins/femtosecond laser lithography/intelligent perception/BP neural network引用本文复制引用
基金项目
装备预研项目(2021-0222/YYQT0222)
东方红航天预研项目(2022330501000146K)
出版年
2024
NETL
NSTL
万方数据