论文部分内容阅读
为补偿捷联姿态测量系统中光纤陀螺因外界干扰引起的高频噪声和强漂移,提出一种基于第二代小波变换和灰色Elman神经网络融合的误差建模和补偿方法。采用Allan方差法分析了在外界干扰下的光纤陀螺输出信号,利用第二代提升小波单独重构的方法分离出陀螺误差模型中的漂移误差和白噪声,灰化漂移误差数据后建立了Elman神经网络模型并进行了补偿。实验结果表明,相较于传统的灰色理论模型和单一的Elman神经网络模型,新算法有效滤除了白噪声,并将预测模型的精度提高到96%以上,证实了模型的有效性。
In order to compensate the high frequency noise and strong drift caused by the external interference in the FOG attitude measurement system, an error modeling and compensation method based on the fusion of the second generation wavelet transform and gray Elman neural network is proposed. The Allan variance method was used to analyze the output signal of the fiber optic gyroscope (FOG) under external interference. The drift error and white noise of the gyro error model were separated by using the second reconstruction wavelet alone reconstruction method. After the drift error data were grayed out, the Elman neural The network model is compensated. The experimental results show that compared with the traditional gray theory model and single Elman neural network model, the new algorithm effectively filters out white noise and improves the accuracy of the prediction model to over 96%, which proves the validity of the model.