基于声音信号的钢材材质检测及试验研究

铁塔加工过程中需要对角钢进行夹紧、冲孔、剪切等操作,不同材质的角钢加工时会发出不同的声音,因此提出利用声音信号检测Q235和Q345两种常用角钢材质。利用传感器和采集仪搭建试验系统,采集声音信号的连续波形,并提取单个冲孔周期声音信号的波形作为后续判别的依据。针对单个周期的声音信号,提取美尔频率倒谱参数(MFCC)作为特征参数,并利用动态时间规整(DTW)方法计算待测模板与Q235和Q345两种标准模板之间的距离,距离小者判定为该种角钢材质。试验分别采集了4种型号角钢的2组样本,验证了上述方法识别材质的有效性。

Steel material detection based on sound signal and the experimental research

In the machining process of iron tower, the angle steel needs to be operated through clamping, punching, shearing, etc, and different material produces different sound, so it is advised to use sound signal to detect steel materials Q235 and Q345. Experiment system is constructed by using the sound sensor and signal acquisition system, so the continuous sound signal is acquired, and the sound wave of single punching cycle is exacted for the follow-up judgement. For the sound signal in one cycle, MFCC(Mel Frequency Cepstrum Coefficient) parameters are extracted as characteristic parameters, and the distances between the being measured template and the two standard templates of Q235 and Q345 are computed by the DTW (Dynamic Time Warping) method. As a result, which distance is smaller determines the template. In the experiment, two sets of samples for four kinds of angle steels are acquired, respectively, which proves the validity of the method for steel material detection.