不完全变形监测数据处理与优化研究

针对指数平滑法和EM算法来预测不完全变形监测数据的潜在值, 分析了两种方法的局限性, 提出了指数平滑法动态选择衰减因子的改进方法和EM算法与切比雪夫多项式组合分析方法, 对比分析结果表明, 采用两种优化方法来生成缺失数据的潜在值均能满足精度要求, 且优化指数平滑法在沉降监测预报的效果更佳。

Processing and Optimization of Incomplete Deformation Monitoring Data

The exponential smoothing method and EM algorithm were used to predict the potential value of incomplete deformation monitoring data. Based on the analysis of limitations of both methods, it was proposed that the exponential smoothing method could be taken as an optimized method for selecting attenuation factor dynamically, and EM algorithm and Chebyshev Polynomial together as a combined analytical method. The comparison of analysis results showed that the potential value of the missing data, which was generated from both methods, can meet the accuracy requirement. It is also concluded that the optimized exponential smoothing method is most effective for monitoring and predicting the subsidence.