大坝监测数据自回归模型特征根的应用研究

Application of characteristic polynomial roots of autoregression time-series model in analysis of dam observation data

为了通过对大坝监测数据的分析来确定大坝安全监测的安全监控指标,应用时间序列分析方法,采用Levin-son递推方法进行参数估计,对大坝监测数据回归分析模型的残差建立AR(p)模型.根据AR(p)模型特征多项式的根与单位圆距离的变化情况来判断大坝结构性态的变化趋势,并据此进行大坝的安全评价.在应用实例中,拱坝在运行过程中产生的裂缝和对坝体所进行的灌浆加固措施,对坝体结构性态所产生的影响以及变化均能够在分析结果中较好地反映出来,表明在大坝观测数据分析时,可以应用该方法有效地进行大坝结构性态的安全评价.

The AR（p） model was applied to analyze the residual data produced by the regression analysis method based on the time series analysis method in order to define the dam safety monitoring index according to the analysis of dam observational data. The Levinson derivation method was used to estimate the AR（p） model parameters. The dam safety behavior trend was appraised according to the distance from the characteristic polynomial roots of AR（p） model to unit circle. An application example of practical deformation data on an arch dam was introduced. The effects of the dehiscence in arch dam and dam body grouting on the dam structure behavior accorded with the calculation results. The method can be effectively used in the safety evaluation of dam structure.