Zheng Gang, Song Linzheng, Xue Wenqi, Zhang Zhiyu, Zhang Benniu
State Key Laboratory of Mountain Bridge and Tunnel Engineering, Chongqing Jiaotong University, Chongqing 400074, China.
School of Civil Engineering, Chongqing Jiaotong University, Chongqing 400074, China.
Materials (Basel). 2024 May 3;17(9):2147. doi: 10.3390/ma17092147.
Coda waves are highly sensitive to changes in medium properties and can serve as a tool for structural health monitoring (SHM). However, high sensitivity also makes them susceptible to noise, leading to excessive dispersion of monitoring results. In this paper, a coda wave multi-feature extraction method is proposed, in which three parameters, the time shift, the time stretch, and the amplitude variation of the wave trains within the time window, are totally derived. These three parameters are each mapped to the temperature variations of concrete beams, and then combined together with their optimal weight coefficients to give a best-fitted temperature-multi-parameter relationship that has the smallest errors. Coda wave signals were collected from an ultrasonic experiment on concrete beams within an environmental temperature range of 14 °C~21 °C to verify the effectiveness of the proposed method. The results indicate that the combination of multi-features derived from coda wave signals to quantify the medium temperature is feasible. Compared to the relationship established by a single parameter, the goodness-of-fit is improved. During identification, the method effectively reduces the dispersion of identification errors and mitigates the impact of noise interference on structural state assessment. Both the identification accuracy and stability are improved by more than 50%, and the order of magnitude of the identification accuracy is improved from 1 °C to 0.1 °C.
尾波对介质特性的变化高度敏感，可作为结构健康监测（SHM）的一种工具。然而，高灵敏度也使它们易受噪声影响，导致监测结果过度离散。本文提出了一种尾波多特征提取方法，该方法完全推导了时间窗口内波列的三个参数，即时移、时间拉伸和振幅变化。将这三个参数分别映射到混凝土梁的温度变化上，然后结合它们的最优权重系数，得到误差最小的最佳拟合温度-多参数关系。在14℃至21℃的环境温度范围内，通过对混凝土梁进行超声实验采集尾波信号，以验证所提方法的有效性。结果表明，利用尾波信号导出的多特征组合来量化介质温度是可行的。与单参数建立的关系相比，拟合优度得到了提高。在识别过程中，该方法有效降低了识别误差的离散度，减轻了噪声干扰对结构状态评估的影响。识别精度和稳定性均提高了50%以上，识别精度的数量级从1℃提高到0.1℃。
