论文部分内容阅读
建立光纤陀螺随机漂移模型以便在滤波中加以补偿是提高光纤陀螺输出精度的有效方法。针对传统光纤陀螺随机漂移建模均采用离线形式,需预先处理数据,不具备普适性等问题,提出一种实时的建模滤波方法。首先,根据大量实测数据对传统离线模型进行改进,研究了一种基于AR模型的在线建立光纤陀螺随机漂移模型的方法。然后,比较了传统Kalman滤波器与H∞滤波器用于实时滤波的效果。实验结果表明,改进型AR模型拟合精度高、普适性强,单个噪声拟合精度最低值为91.6%。H∞滤波器效果优于传统的Kalman滤波器,分析单个噪声滤波效果时,H∞滤波器较Kalman滤波器性能最多可提高38.5%。
The establishment of a fiber optic gyro random drift model to be compensated in the filter is an effective way to improve the output accuracy of FOG. In order to solve the problem that traditional fiber optic gyroscope random drifts modeling is offline, we need to deal with the data in advance and do not have the universality. This paper presents a real-time modeling filtering method. First of all, based on a large number of measured data to improve the traditional off-line model, a method based on the AR model to establish an optical fiber gyro stochastic drift model is studied. Then, the effect of traditional Kalman filter and H∞ filter for real-time filtering is compared. Experimental results show that the improved AR model has high fitting precision and universality, and the lowest single noise fitting accuracy is 91.6%. The H∞ filter is better than the traditional Kalman filter. When analyzing the effect of single noise filter, the performance of the H∞ filter can be increased by up to 38.5% compared with the Kalman filter.