改进的卡尔曼滤波方法在矿区变形数据处理中的应用

Application of Improved Kalman Filtering for Deformation Monitoring in Mine Area

针对卡尔曼滤波方法在处理矿区变形数据中的跟踪能力不强、滤波结果与相对真实值有一定误差的缺点,提出一种基于BP神经网络改进的卡尔曼滤波方法,对滤波结果进行校正,从而增强卡尔曼滤波的跟踪能力,使滤波结果更加接近相对真实值。通过实例计算和分析,修正结果的精度与标准的卡尔曼滤波模型和自适应卡尔曼滤波模型的滤波精度相比得到了很大提高,为高精度的变形预报预警提供了有利条件。

As Kalman filtering did not have very strong capacity for tracking the deformation monitoring data, the improved Kalman filtering based on BP neural network was proposed. The capacity for tracking deformation monitoring data of Kahnan filter was enhanced and the filtered results became closer to the true value,when the results were corrected by BP neural network. The example shows that the accuracy of corrected results higher than the accuracy based on standard Kalman filtering model and the adaptive Kahnan filtering model, and the proposed model provides a high-precision forecasting and warning model for deformation monitoring.