基于 NAR 动态神经网络的石英挠性加速度表参数建模与预测

摘要

Based on accelerometer bias and scale factor of long - term measurement and research environ-ment found in storage at room temperature,the changes of data is highly non - linear,using traditional bi-as and scale factor modeling approach for the long - term the forecast can not achieve good results,this article will NAR model is introduced to model bias and scale factor to go,given the detailed modeling methods and procedures,forecast by the model and the actual value of the phase comparison of error is small,fully shows the effectiveness of the model. Can be studied variation of accelerometer bias and scale factor of the model adopted,and practical applications in accelerometer provides reference for error com-pensation.%通过对石英加速度表零偏和标度因数在常温贮藏环境下长期测量的研究发现，其数据变化具有很强的非线性，利用传统的零偏和标度因数的建模方法，对于长期的预测不能达到很好的效果。将 NAR 模型引入对零偏和标度因数的模型建立中，给出了详细的建模方法和步骤，通过所建立模型进行预测，并与实际值相比较，误差较小，充分说明了模型的有效性。通过所建立模型可以对加速度表零偏和标度因数的变化规律进行研究，并在加速度表实际应用中为其误差补偿提供参考。