Research On Monitoring Intellectualization And System For Resistance Spot Welding Quality

Resistance spot welding (RSW) is one of the most widely used processes in sheet metal fabrication because of its short operation time, low thermal effect, and suitability for robotic automation. To produce a spot weld, electrodes are applied to overlapping region of two pieces of metal, producing a single-lap joint.In order to minimize the number of spot welds and satisfy essential factors such as strength, weld quality must be obtained. Traditionally, to check weld quality, destructive and nondestructive tests were used on randomly sampled workpieces at the production site. As a destructive method, the tensile-shear strength is inspected directly, using samples taken at periodic intervals. As a nondestructive method, spots requiring weld strength are examined visually and by using ultrasonic waves and acoustic emission. However, such conventional methods can only be examined off-line and ensure only the quality of specific random spot samples, having limitation in the widely used assembly line system on which products are continually produced. Weld quality estimation must be done in real time to monitor and repair weld defects as they occur. Over the years many quality estimation systems have been developed by several researchers in order to ensure welds are of high quality.In this paper, electrical and mechanical measurands, such as welding current, electrode voltage, dynamic resistance and electrode force, were monitored for weld quality estimation of SUS304 stainless steel (0Cr18Ni9). To estimate weld quality, a regression model formulated through multiple linear regression analyses and a neural network model were built.Based on advanced sensing and sampling technologies, a data acquisition system for monitoring the dynamic signals was established. The welding current was measured by a Rogowski toroidal coil attached to the lower electrode, then was integrated by integration circuit. In order to increase the system safety, electrode voltage between two tips was isolated by isolation amplifier. A strain sensor, which produced an electric charge proportional to electrode force, was used to monitor the dynamic electrode force. A charge amplifier was used to convert the charge signal from strain sensor into an output voltage proportional to the mechanical input quality. Welding current, electrode voltage and dynamic electrode force were sampled by A/D card at a sampling rate of 10 kHz. The data acquisition system was reliable and stable, suitable for industrial production.An integrated and intelligent on-line welding monitoring system was developed. The system has functions such as on-line monitoring, displaying data and curves, storing and inquiring welding information, which can serve for mechanized and automatic welding production.A series of experiments were conducted to research how process conditions affect the weld quality. The process conditions studied in this research are chosen to be the most often observed in production, such as variations of welding parameters, edge weld, small weld spacing, poor fitup and axial misalignment. The results show that welding parameters are most important for weld quality, and these abnormal process conditions generally lead to a less robust process such as weld expulsion.The characteristics of welding current, electrode voltage, dynamic resistance and electrode force were researched in this paper. The dynamic resistance for stainless steel decreased rapidly at the beginning and then decreased at a reducing rate. As welding current increased, the rate of resistance decrease in the first two cycles increased. Also, as the current increased, there was a decrease in the resistance level. Lower electrode force induced a general increase in the resistance level. During the weld stage, the actual electrode force was larger than the preset one. The change of dynamic electrode force was caused by thermal expansion and the yielding of the weld area at a high temperature. Several factors were extracted for weld quality estimation based on the dynamic resistance and electrode force pattern. The correlation between these factors and nugget diameter were observed through the regression analysis, and r3, r8, f2, f3 and f6 were used as the weld quality factors.Expulsion is the common defect in resistance spot welding, it is therefore important to monitor or predict expulsion so that modifications to welding parameters or other remedial actions can be initiated to reduce the incidence of expulsion. The results show that when an expulsion occurred, the dynamic resistances and electrode voltage signal experienced a sudden drop and dynamic electrode force signal had a vibration signature. Thus monitoring the signals of electrode voltage, dynamic resistance and electrode force can give an indication of whether there is expulsion in the finished weld.To estimate weld quality for resistance spot welding, multiple linear regression analyses was used. In the quality estimation model, weld quality estimation was influenced by welding current I, welding time T, the endpoint value of resistance curve r3 and electrode force increase f3. The regression model error is high when there is no nugget, and the maximum error was only 10.2% for these weld spots with nuggets.A BP neural network was built to estimate weld quality, which showing accurate results in nugget diameter estimation. The neural network had fine fault tolerance and the maximum error was only 5.6%, higher than the linear regression model. According to the results described in this paper, the real-time quality estimation is made possible.