【作者】 陈凯；

【导师】 陈善本； 陈华斌；

【作者基本信息】 上海交通大学 ， 材料科学与工程， 2019， 硕士

【摘要】 焊接电参数,包括焊接电流,焊接电压等反应了焊接电源的动态特性。而焊接声音也从另一个角度提供了丰富的焊接动态信息。两者的结合可以在一定程度上更为全面地反应焊接动态过程及焊接质量信息。为了开展这一研究,设计并进行了低碳钢薄板的CMT(Cold Metal Transfer)搭接焊实验。CMT焊接方法具有无飞溅、低热输入的特点,广泛用于薄板焊接、异种金属焊接等领域。本文首先研究了CMT的声源特性。CMT的不同操作模式具有不同的声音特性。其中,进一步对“特殊两步模式”的焊接声音分析,得到结论:焊接电弧声信号来源于焊接电弧的能量变化,并且电弧能量变化越快,对应的声压值越大。在此基础上,对CMT焊接的异常状态信号采集方法开展了研究。共开展了两组实验。第一组在其他工艺参数不变的条件下,调节气流量,直至停止送气,模拟了送气故障;第二组实验调整搭接间隙至2mm时,得到了焊漏的焊接异常状态。进一步研究了焊接电参数与焊接声音信号的特征提取和多传感信息融合方法。在电信号方面,提取了焊接过程的焊接电流,焊接电压,线能量等;在声音信号方面,声音信号经过分帧,加窗后提取了Mel倒谱参数MFCC。应用BiLSTM-CTC算法可以识别焊漏,送气故障等焊接过程信息。对送气状态和焊漏这两类异常焊接状态建立的识别模型,基于声音特征的分类错误率最低为0.389,基于电信号特征的分类的错误率最低达到了0.774,而基于融合特征的分类错误率为0.295。这些实验证明,在同一个模型框架下,融合了焊接电信号特征与声音信号特征的模型表现出的分类错误率最低。为了达到实时监测的目标,本文基于B/S模式开发了实时监测系统。这一系统应用了前文的分析模型,能够基于实时的焊接电信号及声音信号判断CMT搭接焊接的稳定性,识别焊接异常状态。更多还原

【Abstract】 Welding electrical parameters,including welding current and welding voltage,reflect the dynamic characteristics of welding power source.Welding sound also provides abundant dynamic information from another point of view.The combination of the two methods can reflect the dynamic welding process and welding quality information more comprehensively to a certain extent.In order to carry out this research,CMT(Cold Metal Transfer)lap welding experiments of low carbon steel sheet were carried out.CMT welding method has the advantages of no spatter and low heat input.It is widely used in thin plate welding,dissimilar metal welding and other fields.Firstly,the sound source characteristics of CMT are studied.Different operation modes of CMT have different sound characteristics.Further analysis of the "special two-step mode" of welding sound shows that the acoustic signal of welding arc originates from the energy change of welding arc,and the faster the change of arc energy,the greater the corresponding sound pressure value.On this basis,the signal acquisition methods of abnormal CMT welding status were studied.Two groups of experiments were carried out.In the first group,the gas flow was adjusted until the gas supply stopped under the condition of other technological parameters unchanged;in the second group when the lap gap was adjusted to 2 mm,the defect of welding wear achieved.Furthermore,the feature extraction and fusion methods of welding electrical parameters and welding sound signals were investigated.In the aspect of electric signal,welding current,welding voltage,line energy and so on;In the aspect of sound signal,MFCC is extracted after de-framing and windowing.BiLSTM-CTC algorithm can be used to identify welding process information such as welding wear,gas feeding fault and so on.For the recognition model of gas supply status and welding wear defects,the classification error rate based on sound feature is the lowest at 0.389,the classification error rate based on electrical signal feature is the lowest at 0.774,and the classification error rate based on fusion feature is at 0.295.These experiments prove that under the same model framework,the model which combines the characteristics of welding electrical signal and acoustic signal has the lowest classification error rate.In order to achieve the goal of real-time monitoring,this paper develops a real-time monitoring system based on B/S mode.This system applies the analysis model mentioned above,and can judge the stability of CMT lap welding and identify abnormal conditions based on real-time welding electrical and acoustic signals.更多还原