基于卡尔曼滤波与AR混合算法的沉降监测

Deformation Prediction Modal Combined with Calman and AR Algorithm

针对非平稳序列观测数据,结合卡尔曼滤波递推算法等的优点,建立了基于卡尔曼滤波的AR自回归模型。此算法不需象常规算法那样提取趋势项,而是在状态方程中设置AR参数方差的随机模型,利用观测数据建立观测方程。模型应用于输水隧洞开挖过程的的沉降监测,取得了较好的效果,说明此方法准确、可行。

The non-stationary time series data were deaIt with the hybrid algorithm modal which combining the combined Caiman and AR algorithm, and the modal was build, in which the state equations was set with the parameters of stochastic variance of the AR modal instead of extracting the tendency items from the original data in conventional AR modals, and measure equations were constructed with the observation data, then the AR modal can be solved by the Caiman algorithm. The deformation prediction results of the modal used in water transportation tunnel construction showed that this method was accuracy and feasible.