| Resistance spot welding is one of main welding methods in the manufacturing process of stainless steel urban railway passenger cars,and the quality of resistance spot welding has an important influence on the safety of the body structure.The process of spot welding is a comprehensive result of heat,electricity,power and other physical fields.The formation of nugget is completely closed,which makes it difficult to evaluate the quality of resistance spot welding.Under conditions of comprehensively evaluating the quality of resistance spot welding joints and ensuring product performance,the proportion of the additional spot welding joints is reduced,so that a huge production costs of spot welding can be saved.At present,many researchers have made fruitful results in the spot welding quality assessment,but most of them are still in the experimental stage.The problem of comprehensive evaluation spot welding quality in workshop has not been satisfactorily resolved.The main limitation is the lack of reasonable,efficient and economical means of spot welding quality assessment.In this paper,the dynamic resistance signal in the process of resistance spot welding and the ultrasonic testing signal are combined,the tensile strength of the spot welding joint is used as the criterion for the quality of spot welding,and the comprehensive evaluation of SUS301-HT stainless steel double pulse resistance spot welding quality is first postulated.Based on the monitored welding current,electrode voltage and electrode pressure signals,the dynamic resistance curves of stainless steel resistance spot welding under expulsion and wornelectrode were studied,respectively.The effects of welding parameters under expulsion condition on dynamic resistance were studied.The results show that the dynamic resistance curve is similar,the curve distribution range is narrow,and the fluctuation is small when the workpiece surface is clean.The number of instantaneous dynamic resistance drops was perfect match with the number of expulsions occurring.The number of expulsion,the time of first expulsion and the magnitude of dynamic resistance were proposed to characterize the severity of expulsion,which give some guidance on the quality of resistance spot welding.To study the relationship between electrode wearing and dynamic resistance curves,electrode tips with different sizes were used to simulate electrode wearing.Due to a certain degree dispersion of the dynamic resistance curve,the slight change of the electrode tip size may not affect the dynamic resistance.the average dynamic resistance of the multiple spot welding joints is proposed to represent the dynamic resistance curve under certain process parameters.It is shown that the partial resistance values of average dynamic resistance decrease with the size of electrode tip increasing.The correlation analysis about the dynamic resistance of the double-pulse stainless steel resistance spot welding and the tensile shear strength of the spot welding joint shows that the dynamic resistance curve during the preheating period is less correlated with the tensile shear strength of the spot welding joint,while the dynamic resistance curve during the secondary pulse except the maximum contact resistance is obviously correlated with the tensile shear strength of the spot welding joint,which can be used as the characteristic parameters for quality evaluation model of stainless steel resistance spot welding.An intelligent evaluation model for classifying spot welding joints and suspicious spot welding joints is established basing on the fact that the process parameters used in the production process have been tested for many times which means that most of the spot welding spots are qualitied welding spots,and under the same welding parameters the dynamic resistance curves are similar.Based on the off-line clustering model classifying the welding joints,an online clustering algorithm named Leader-Follower is used,which makes up the limitation of off-line clustering model having fixed the number of classification.However,the number of clusters classified by the basic Leader-Follower is big,which leads to poor readability of the classification results and is bad for automatic identification of qualified spot welding joints.The improved Leader-Follower algorithm of online clustering considering the update and fusion of classes is immune to the order of samples.The results show that the improved Leader-Follower algorithm can classify the samples correctly except for a small number of samples during the process of creating a new class and is robust to sample order.The ultrasonic signal is mainly used to detect the nugget and shrinkage.However,the problem is how to get high signal noise ratio(SNR)characteristic signals.A new method is proposed to reconstruct wavelet packet coefficient.The ultrasonic A scanning signal is decomposed by the wavelet packet,the ratio of the variance in the sampling interval between each node and the characteristics of the nugget are obtained which are taken as the coefficients of the nodes,and the A-scanning signal with high SNR is reconstructed by the new coefficients.The size of shrinkage calculated by this method agrees very well with metallographic measurement.Compared with the time domain and frequency domain methods,the proposed algorithm is less affected by the workpiece and has high precision.At present,the method of assessing the quality of resistance spot welding with ultrasonic signal is mainly to identify the nugget characteristics by interface reflection wave method.However,the "nugget" area calculated by this method is generally different from the actual nugget size which is negatively correlated with the tensile strength of the spot weld,when the indentation is deep.In this paper,the wavelet decomposition function is designed according to the frequency and bandwidth parameters of the ultrasonic probe,and the effective nugget area is extracted by wavelet principle.The experimental results show that the method has obvious advantages over the existing methods.The relationship between the effective nugget area and the tensile shear strength of the spot weld can be described by the linear regression model.A high efficiency and economical method combining dynamic resistance curve signal and ultrasonic testing signal for stainless-steel resistance spot welding quality evaluation is proposed.The establishment of the model is carried out simultaneously with the evaluation of the spot welding quality.The dynamic resistance characteristics and the tensile shear strength are taken as the input and output,respectively.In the process of model learning,the evaluation results of the tensile strength by ultrasonic signal are regarded as labels.Some spot welding joints must be detected by ultrasonic signals when they cannot be assessed by quality evaluation model based on dynamic resistance.The proposed method not only be able to exert their advantages but also to overcome the drawbacks of low examination efficiency in single element ultrasonic testing and to solve the limitation of establishing the quality model depending on the relevant empirical parameters only based on the dynamic resistance characteristic.In summary,the following innovative results in the field of resistance spot welding quality assessment are achieved: The problem of shrinkage quantitative detection by ultrasonic signal is solved,and the effective nugget for evaluating the tensile shear strength is extracted,which establish the foundations for the comprehensive evaluation of the spot welding quality.The proposed method combining dynamic resistance signal with the ultrasonic signal achieves highly efficient,economical spot welding quality evaluation. |
PDF Full Text Request