利用声发射信号时频特征在线监测慢走丝线切割加工表面粗糙度

摘要

Aiming at the problem that monitoring of Wire Electrical Discharging Machining-Low Speed (WEDM-LS) process is difficult because of high temperature, narrow kerf and other factors, Acoustic Emission (AE) was adopted to implement on-line monitoring of WEDM-LS process with adjustable duty ratio pulse.Firstly, AE signal was decomposed into eight independent frequency bands by energy spectrum algorithm of wavelet packet, respectively being W1~W8 and frequency reduces in sequence;then energy characteristic of all frequency bands was extracted to research the relativity between it and surface roughness value of machining work-piece.Experimental results show that energy of W8 frequency band is highly related to surface roughness value and energy of this frequency band and pulse discharge energy increase with increase of pulse duty ratio, and roughness value of machining surface increases with it gradually.Finally, a mathematical prediction model between the surface roughness value and the energy in W8 frequency band was established via regression analysis, and error was only 3.51% between prediction result and surface roughness actually measured.It can be illustrated that this model has high prediction precision for online monitoring of machining surface roughness.%针对慢走丝线切割加工(WEDM-LS)时因温度高、切缝窄等因素造成的加工过程难以监测的问题,利用声发射检测技术对占空比可调脉冲的慢走丝线切割加工过程进行在线监测.首先利用小波包能谱算法将AE信号分解到8个独立的频段上:分别为W1~W8,且频率依次降低;然后提取各频段上的能量特征,研究其与加工工件表面粗糙度值之间的相关性.试验结果表明:W8频段的能量与表面粗糙度值之间具有较高的相关性,该频段的能量与脉冲放电能量均随着脉冲信号占空比的增大而增大,且加工表面粗糙度值也随之逐渐增大.最后通过回归分析得到了反应材料表面粗糙度值与W8频段能量占比关系的数学预测模型,该模型的预测结果与实际测得的表面粗糙度值误差仅为3.51%.说明该模型具有较高的预测精度,可用于加工表面粗糙度的在线监测.