监测位移预报的非等间距相关系数平稳序列法

建立了监测位移时间序列分析、建模和预报的非等间距相关系数平稳序列法.采用随机过程理论导出了模型的整体极大似然估计,在保证参数估计量优良统计特性的同时,克服了现行组合模型方法难以处理非等间距监测位移序列的问题.通过数字模拟仿真技术解释了传统插值处理产生较大误差的原因.计算实例表明该方法具有较高的建模和预报精度.

Prediction method of monitoring displacement based on correlation coefficient stationary series with unequally spaced data

An analysis method of unequally spaced correlation coefficient stationary series was established, which can be used into model buildings and prediction of displacement monitoring in geotechnical and hydraulic engineering. The integral maximum likelihood estimations of model parameters were derived from stochastic process theory; it can not only guarantee the good statistical properties but also solve the problem of unequally spaced time series analysis. Traditional interpolation method will result in an unacceptable error when the rate of data missing was relatively high, which was proved by the Monte Carlo simulation. It is shown that the estimations of the variance of displacement series are far from the true value. An example was given in the last section, which illustrates the advantage of presented method.