干涉型光纤陀螺仪零漂建模方法

Modeling technology for null drift of FOG

光纤陀螺仪零漂是衡量光纤陀螺仪精度的重要指标。文中对某光纤陀螺仪的零偏数据进行分析和预处理,采用时间序列分析法建立了AR(2)模型,同时基于BP神经网络建立了预测模型。建模结果分析表明:AR建模方法简单易行,但适用性不如BP网络模型,BP网络模型建模精度高,但算法复杂且收敛速度慢,容易陷入局部极小,因此采用了改进训练方法,改善了BP网络性能。

The null drift is an important index to measure the precision of FOG In this paper, the bias data of FOG were analyzed and preprocessed, and the time sequence method was used to establish the AR（2） model. Meanwhile, the predicting model was established based on BP ANN. The modeling results show that the AR modeling method is simple and feasible, but it is not as adaptable as BP ANN model. The BP ANN model has high precision hut the algorithm is complex. It also has problems of local minimum and slowly converging, so an improved training method was used to improve the performance of BP ANN.