Abstract:

As the mathematical model that is built in traditional Kalman filtering is not very accurate and its statistical characteristics of dynamic noise are difficult to confirm, the traditional Kalman filtering may lead to filtering divergence, even may lead to evaluation distortion. Therefore, the method of selfadapting Kalman filtering is put forward to solve this problem. In this article, traditional Kalman filtering, selfadapting Kalman filtering based on maximum a posterior (MAP) estimation principle, and selfadapting Kalman filtering based on variance compensation are used to process the data in settlement prediction of Metro tunnels, and analysis and comparison are made among these three methods. The result shows that, compared to the other two methods, the method of selfadapting Kalman filtering based on variance compensation is more accurate in settlement prediction. 

Key words: Kalman filtering, variance compensation, maximum a posterior (MAP) estimation, selfadapting Kalman filtering, data processing