基于吸引子特征量的非线性判定法在大坝位移监测数据时间序列中的应用

Application of Nonlinear Decision Method Based on Attractor Characteristic Quantity in Time Series of Dam Displacement Monitoring Data

摘要: 由于大坝位移监测数据存在一定的噪声，为了解决高噪声水平下非线性判定的准确性问题，引入基于吸引子特征量的非线性判定方法。本文对比分析了Lorenz方程和Henon映射2种非线性混沌时间序列在不同噪声干扰情况下，时间反演不可逆量及吸引子特征量分别作为特征统计量时替代数据法的检验性能，发现以Hurst指数作为非线性特征量时的替代数据法显示出较强的鲁棒性。将其应用于实际工程中的大坝三个测点的位移监测数据时间序列，结果表明大坝监测数据位移时间序列呈现出非线性变化的特点，为分析大坝结构形态的变化，及时掌握大坝的运行状态奠定基础。

Abstract: Because of the noise of dam displacement monitoring data, a non-linear judgment method based on attractor characteristic is introduced to solve the accuracy of non-linear judgment under high noise level. In this paper, the test performance of two kinds of nonlinear chaotic time series, Lorenz equation and Henon map, is compared and analyzed when the time inversion irreversibility and attractor characteristic are used as feature statistics respectively under different noise dis-turbances. It is considered that the surrogate data method with Hurst exponent as the non-linear characteristic shows strong robustness. It is applied to the time series of displacement monitoring data of three dam observation points in practical engineering. The results show that the dis-placement time series of dam monitoring data presents the characteristics of non-linear change. This lays the foundation for analyzing the change of dam structure and grasping the operation state of the dam in time.