摘要
针对声波反射法检测基桩质量的应用,提出了一种智能提取模型.该模型首先选取标准样本信息对系统进行训练,并提取适合于测量环境的模型参数;然后将参数输入到由FFT变换、小波去噪、Hilbert变换组成的混合模型中,提取由反射波引起的瞬时相位突变点,实现对被测基桩的缺陷位置和长度的检测.将该模型应用于水电站锚杆锚固质量检测,实测结果为:平均长度测量误差和平均缺陷定位误差分别为2.5%和2.1%,大大优于常规方法,且能在现场实时给出测量结果.
An intellectual extraction model for the engineering evaluation of pile testing was developed. First, the method selects standard sample to train the system and extract module parameters suitable for the measuring environment, then, inputs these parameters into the hybrid model based on the FFT analysis, wavelet de-noising and Hilbert transform to extract the instantaneous phase change^point caused by reflected wave. The method was applied to the bolt anchoring quality inspection on a Hydropower Station. The experimental results show that the model can output measure results in real-time automatically, and the average relative error for detecting length and defect location are 2.5% and 2.1% in turn, which is much better than the conventional method.
出处
《地球物理学进展》
CSCD
北大核心
2013年第1期523-530,共8页
Progress in Geophysics
基金
国家自然科学基金专项基金项目(40727001
40774073)
国家973项目(2007CB209607)联合资助
