【摘要】 大坝安全监测数据是既可能有趋势性又可能有季节性的时间序列数据,为了准确地对其进行分析与预测,运用较适合处理非平稳性时间序列的ARIMA模型,首先通过差分将非平稳时间序列平稳化,然后通过自相关函数和偏自相关函数进行模型识别,得到若干初选模型,进而估计各初选模型中的参数,并根据贝叶斯信息准则确定最终模型,最后利用最终模型对监测数据进行拟合和预测。以李家峡水库大坝的一组安全监测数据为例,通过模型计算,对模型拟合和预测数据与实际位移数据进行比较,结果表明,ARIMA模型在大坝安全监测数据的分析与短期预测方面有较高的精度,具有可行性。更多还原

【Abstract】 Dam safety monitoring data is time series data with trend and seasonal characteristics. In order to analyze and predict it accurately,ARIMA model could be applied to deal with this unstable time series. In details,firstly,it got stable time series by difference. Secondly,it identified the model by AFC and PAFC and got some trial models. Thirdly,it estimated the parameters of these trial models and confirmed the final model according to Bayesian Information Criterion. Finally,it fitted and predicted the monitoring data with the final model. Taking Lijiaxia dam as an example,it built the model based on above process and compared the model data with monitoring data. The result shows that ARIMA is accurate when analyzing dam safety monitoring data and predicting it in a short-term. In short,ARIMA is a feasibility study.更多还原