基于声发射的木材裂缝数量辨识研究

Research on the Number of Crack in Wood Based on Acoustic Emission

针对木材裂缝缺陷,提出一种基于声发射的木材裂缝数量识别方法。首先,在试件上人为依次制作1 mm×9 mm(长×高)的4条裂缝,在裂缝的一侧通过折铅的方式产生声发射(acoustic emission,AE)信号,另一侧放置传感器,信号采样频率设置为2 MHz。然后,通过粒子群算法(particle swarm optimization,PSO)确定变分模态分解(variational mode decoposition,VMD)的分解层数K和惩罚因子α,并将原始信号分解为具有不同频率的本征模态(intrinsic mode function,IMF)。接着,随机选择5组信号进行VMD分解,并对分解后的IMF构成的矩阵进行奇异值分解(singular value decomposition,SVD),得到相应的奇异值向量,再由5组奇异值向量组成标准矩阵。最后,由测得的AE信号,分别与标准矩阵计算马氏距离,并依据最小判别原则,判定裂缝数量。结果表明,PSO-VMD-SVD方法能够方便提取出AE信号特征,并通过计算马氏距离进行裂缝数量判别,判别正确率为92%。

A method for identifying the number of wood cracks based on acoustic emission is proposed for wood crack distances.Firstly,four cracks of 1 mm×9 mm(length×height) are artificially made in sequence on the specimen,and an acousitic emission(AE) signal is generated on one side of the crack by folding the lead,and the sensor is placed on the other side with a signal sampling frequency set to 2 MHz.Then,the number of decomposition layers K and the penalty factor α of the variational modal decomposition(VMD) are determined by the particle swarm algorithm(PSO),and the original signals are decomposed into the intrinsic mode function(IMF) with different frequencies.tFive groups of signals are then randomly selected for VMD decomposition,and the matrix composed of the decomposed IMFs is subjected to singular value decomposition(SVD) to obtain the corresponding singular value vectors,and then the standard matrix is composed of the five groups of singular value vectors.Finally,from the measured AE signals,the Mahalanobis distance is calculated with the standard matrix,respectively,and the number of cracks is determined based on the principle of minimum discrimination.The results show that the AE signal features can be easily extracted by the PSO-VMD-SVD method and the number of cracks can be discriminated by calculating the Mahalanobis distance,and the correct rate of discrimination is 92%.