BP-Bagging模型在光纤陀螺温度补偿中的应用

Application of BP-Bagging model in temperature compensation for fiber optic gyroscope

为了消除光纤陀螺的温度效应并提高陀螺的精度,BP神经网络模型广泛的应用在光纤陀螺的零偏温度漂移辨识和补偿中。然而,单神经网络模型的泛化能力差,影响模型的预测结果。结合神经网络集成学习的思想,利用Bagging集成技术产生差异大、预测能力强的个体网络,提升模型的预测能力。建立光纤陀螺零偏温度的BP-Bagging模型,将其应用在温度补偿中。通过对某型光纤陀螺的零偏漂移数据进行仿真,结果表明:BP-Bagging模型相比线性回归模型、单BP神经网络模型的补偿效果更显著,有效改善了陀螺的零偏稳定性能。

In order to improve the precision of fiber optic gyroscope (FOG), BP neural networks are widely applied in identification and compensation of FOG bias drift caused by temperature variation. However, the single BP neural network model is poor in generalization ability, which can affect the stability of prediction results. According to the ideas of ensemble learning, a neural network ensemble is developed to effectively generate the individual learner with strong generalization ability and great diversity by using Bagging algorithm, which has higher stability and accuracy in prediction, compared with the single BP model. A BP-Bagging model is established to compensate the FOG temperature errors. The traditional modeling method of linear regression and single BP neural network are also investigated to provide a comparison with the novel proposed model. The simulation results show that the BP-Bagging approach has better performance compared with those traditional models in compensation of FOG temperature drift and improvement of FOG accuracy.