【作者】 胡家琨；

【导师】 高进强；

【作者基本信息】 山东大学 ， 材料加工工程， 2006， 硕士

【摘要】 弧焊机器人CO₂焊接过程存在多种干扰因素，比如工件尺寸的加工和装配误差、工件热变形及弧长变化等等。这些干扰因素对焊接质量有重要影响。在焊接过程当中，如果能够实时地检测到焊接过程的干扰信息，及时调整焊接过程的工艺参数，实现对焊接质量在线监控，则可以避免焊件的报废，保证焊接接头的质量，具有巨大的经济效益和工程实用价值。 电弧焊接过程是一个多因素交互作用的复杂过程，焊接过程的状态与许多参数有关。然而在焊接过程当中，当焊接条件发生改变时，有些参数变化明显，而有些则不甚明显。因而，如何从这些参数中提取出能够明显表征不同工艺条件的特征信息，便成为焊接过程实时监控的重要一环。不同工艺条件下的焊接质量的好坏，肯定会反映在与焊接过程本身固有的焊接工艺参数上，如焊接电流、焊接电压、电弧声压、熔滴过渡频率等等。众所周知，运用电弧传感对焊接过程电参数（焊接电压、焊接电流）进行检测，具有经济性、实用性、可靠性等优点。 本文以对接接头和T型接头为研究对象，开展机器人CO₂焊接工艺试验，采用电弧传感的方法实时测量和处理焊接电参数信号。分别在两种焊接接头中人为地加工出缺口，以模拟实际焊接生产过程中的装配间隙等干扰因素。研发出基于NI-6221数据采集卡和LabVIEW软件的机器人CO₂焊接电参数实时采集和处理系统，实时地提取机器人CO₂焊接过程有干扰时的特征信息。结果表明，两种焊接工艺条件下，所提取出的焊接过程特征信息并不一样，但是均能很好地反映所对应的焊接过程，证实了焊接过程的特征信息与干扰因素的相关性。 针对对接接头，本文设计了基于SPC（统计过程控制）方法的均值控制图，以对接接头焊接过程的特征值U_M作为控制变量。利用正常焊接过程时的U_M处于受控状态，而有扰动焊接时的U_M则处于非受控状态的原理，对焊接过程进行实时监控。当焊接过程处于非受控状态时，监控系统就自动发出警报。结果表明，利用此方法进行实时监控能够达到监控的要求。 针对T型接头，本文开发了模糊Kohonen神经网络系统，以T型接头焊接过更多还原

【Abstract】 There are many kinds of disturbing factors during robotic CO₂ welding, for example, the machining and fitting errors of the workpiece, the thermal distortion of the workpiece and the change of arc length during welding process and so on. These factors have important effects on the weld quality. If the disturbing factors are detected in real time, on-line monitoring of weld quality can be realized through adjusting the welding process parameters in order to avoid the defective welds and assure high weld quality. Thus, it has big economic benefit and engineering practicality.Arc welding process is a very complicated process with many interacting factors mixed together, so the transient state of welding process is related to many parameters. However, when the welding conditions change, some parameters change clearly while other parameters do not. So how to extract obvious feature information from the process parameters, which describes different welding conditions, is very important to realize the real-time monitoring. The weld quality under different welding conditions is related to the welding parameters which are inherent electrical parameters in the welding process itself, i.e. welding current, welding voltage, arc sound, short-circuiting frequency and so on. It is well known that using arc sensors to detect the electrical parameters （welding voltage and welding current） is very economical, practical and reliable. In this study, robotic CO2 welding experiments for butt joint and T joint are carried out. Arc sensor is used to measure and process the electrical parameters during the welding process. At the middle part of the welding seam, a notch is intentionally machined. This sudden increased gap along the welding seam is used to simulate the fitting errors of the joint in practical production. The real-time acquiring and processing system of the measured signals are developed based on更多还原