论文部分内容阅读
激光陀螺漂移时间序列具有非平稳和非线性的特点,单一预测模型难以准确跟踪其变化趋势。研究了基于经验模态分解(EMD)和最小二乘支持向量机(LSSVM)的多尺度混合建模方法及在激光陀螺漂移预测中的应用。首先,利用经验模态分解将漂移时间序列分解为多个本征模式分量,在采用具有适当核函数的最小二乘支持向量机分别对这些分量进行预测后,以加权集成方式得到最终预测结果。最后,将该方法用于激光陀螺的随机漂移预测中,仿真结果表明:该方法能够准确预测激光陀螺漂移值,取得了比单一模型更好的预测效果,能够为激光陀螺的漂移补偿、故障预报和可靠性诊断提供参考。
The laser gyro drift time series has the characteristics of non-stationary and non-linear, and the single prediction model can not accurately track its changing trend. The multi-scale hybrid modeling method based on Empirical Mode Decomposition (EMD) and Least Squares Support Vector Machine (LSSVM) is studied and its application to laser gyro drift prediction is studied. First, we decompose the drift time series into multiple eigenmode components by using empirical mode decomposition. After predicting these components by least square support vector machine with appropriate kernel function, the final prediction results are obtained by weighted integration. Finally, the method is applied to the random drift prediction of laser gyroscope. The simulation results show that this method can accurately predict the laser gyro drift and achieve a better prediction effect than the single model. It can compensate the drift of the laser gyro, And reliability diagnosis provide a reference.