基于时间序列分析的桥梁长期挠度分离与预测

Study on Separation and Forecast of Longterm Deflection Based on Time Series Analysis

摘要:

针对挠度长期监测信号的多尺度特性及温度荷载下挠度周期性变化的特点，利用修正平均周期图法获取信号分量周期长度，结合基于中心移动平均法的挠度温度效应分离策略，提取结构变形特征，运用ARIMA模型建立结构变形趋势的预测模型.以黄冈长江大桥为例，讨论了温度与挠度的相关性，通过模拟数值试验验证分离算法的有效性.实测数据的处理与分析结果表明，本文方法具有较高的可靠性和精度，可为桥梁进行损伤识别提供数据支持.

Abstract:

In view of the multiscale property of deflection and the periodicity under temperature load, a strategy was proposed to separate the deflection and extract the longterm trend, which combines Welch method with Center Moving Averages Method. A prediction model of the deformation was established based on ARIMA model. Taking Bridge of Huanggang Yangtze River as an example to issue the correlation between temperature and deflection, and test of numerical simulation was made to validate the effectiveness of the algorithm. Processing and analysis of the measured data indicate that the approach of this article has high precision and reliability which could provide data support for damage identification of bridge.