基于灰色近似支持向量机的加速度计参数预测

Accelerometer Parameters Prediction Based on Grey Proximal Support Vector Machine

温度、振动等环境载荷使得石英挠性加速度计参数产生漂移,直接影响了惯导系统的测量精度和性能,石英挠性加速度计参数变化趋势为非线性的,很难用常规的建模方法进行趋势预测。基于灰色理论适合进行小样本、贫信息不确定型系统建模以及近似支持向量机不需要求解二次规划就能求得非线性模型参数的优点,提出了基于灰色近似支持向量机进行石英挠性加速度计参数预测的方法。为了验证该方法的有效性,针对自然贮存的加速度计进行了固定周期的参数标定,结果表明灰色近似支持向量机具有很好的预测效果。

Environment factors like temperature and vibration influence parameters drift of quartz flexure accelerometer,which directly affects the accuracy and performance of inertial navigation system.Because the parameters variation trend of quartz flexure accelerometer is nonlinear,it is difficult to forecast it with the conventional methods.As small sample copy and poor information uncertainty problem can be easily solved by grey theory and proximal support vector machine model parameters can well obtained without quadratic programming,an accelerometer parameters prediction method based on the grey proximal support vector machine is proposed.The calibration of natural environmental scale value of accelerometer parameters is carried out to verify the effectiveness of this method.The results show that grey proximal support vector machine has got excellent prediction performance.