摘要
陀螺电机状态直接影响惯导系统的精度和可靠性,对其进行预测是惯导系统性能评估和寿命预测的重要途径。利用灰色理论的建模预测方法对随机性较大的数据预测精度不高;BP神经网络模型的预测方法具有良好的非线性和自学习能力,但训练效率不高且训练效果受样本数影响较大,网络容易限于局部最小值。针对陀螺电机状态特征参数的特点,本文提出一种基于灰色BP神经网络的混合模型。该模型利用BP神经网络对灰色模型误差进行建模,模型输出返回灰色模型进行输入修正。利用灰色理论、BP神经网络以及混合模型对状态特征参数进行建模和预测,结果表明,混合模型的预测误差比灰色模型减小了约2/3,比神经网络减小了约1/3,证明了该模型的有效性。
The accuracy and reliability of inertial navigation system （INS） is directly influenced by gyro motor＇s state. The prediction of which is an important way for INS performance evaluation and life prediction. The prediction accuracy of grey theory is limited by the dam＇s randomness. BP neural could solve nonlinear problem and performs well in self-adaption and self organization, but it＇s training effect and efficiency is limited by the number of samples. In this paper, a hybrid model combining the advantages of grey theory and BP neural network is put forward based on characteristics of gyro motor＇s state parameters, in which grey model error is modeled by BP neural network, and grey model uses it＇s output to correct its input. Grey theory, BP neural network and the hybrid model are constructed respectively to model and predict the state parameters. The results indicate that the prediction errors of hybrid model decrease 2/3 in comparison with that of grey model and decreases 1/3 in comparison with that of BP neural network, which prove the validity of the hybrid model.
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2010年第1期120-123,共4页
Journal of Chinese Inertial Technology
基金
海军装备部重点装备预研基金
