彭德红1,孙德立2
钢筋混凝土梁损伤声发射信号特征提取与识别研究
彭德红1,孙德立2
摘 要:钢筋混凝土桥梁暴露在恶劣工作环境下常常受到裂纹损伤威胁,及时有效地检测到混泥土梁的微裂纹以及识别梁的损伤状况是保证桥梁安全作业的重要基础。针对目前采用全波形声发射仪采集声信号受到噪声污染而影响探伤精度的问题,提出了基于神经网络的非线性独立分量分析(ICA)与维格纳分布(WVD)的损伤声发射信号特征提取与识别的新方法。通过RBF网络估计混入声信号中的非线性噪声成分,利用ICA算法分离出真实的梁损伤声信号,消除噪声干扰,并应用WVD分析分离信号的时频谱分布,提取信号特征频率处的能量特征作为识别损伤状态的有效参数。三点弯曲加载试验结果表明,非线性ICA能够有效抑制噪声导致的声信号频率漂移,得到信号可靠的关键特征,提高梁裂纹损伤的识别精度,且结果比不进行去噪处理提高了9%。
Abstract:Reinforced concrete beams often suffer from cracks due to severe operation conditions. Therefore, it is imperative to monitor the beam condition in time to prevent damages. However, fault acoustic signal recorded by the waveform acoustic emission instrument is submerged by noise. To eliminate the noise, a new approach to identify the beam damages is presented. It firstly uses a RBF neural network to estimate the mixed nonlinear noise in the original acoustic signal, then the independent components analysis (ICA) is used to separate the true fault signal with the noise removed, lastly, the Wigner-Ville distribution (WVD) is adopted to extract distinct energy features of the separated signal in the time-frequency domain to realize the recognition of the beam health states. The three points bending loading test has been carried out to evaluate the proposed detection method. The experiment results show that the mixed noise can be eliminated effectively by the use of nonlinear ICA, and the identification precision is improved. Moreover, the performance of the proposed method is superior to the method without the ICA processing.
版权所有:中国混凝土与水泥制品网 苏ICP备10086386号 网站建设:中企动力 苏州