Arc Welding Quality Evaluation Technology Based On Entropy Measures And Study Of Welding Shop Digital Control System

As one of the most important material forming and processing technologies in modern manufacturing industry,the welding technology is nearly used in all of the industrial manufacturing fields.Welding quality directly relates to the product quality,reliability and service life,and it also relates to production cost,efficiency,market and the like.Thus,the construction of welding quality management system is very necessary.In the situation of the large number of welding equipment and the scattered distribution of them in the arc welding workshop,a networked welding process quality management system with both software and hardware was developed.The network database management of welding procedure and the related information was realized.Meanwhile,the synergetic acquisition,analysis,network transmission and database management of welding process parameters together with the visual image of molten pool was realized too.Because of the droplet transfer,and the complex interaction between two arcs,the factors that can influence the stability of double wire pulse metal inert gas(PMIG)welding are more complex than other arc welding methods.The analysis of welding process parameters and weld pool image are mainly carried out on aluminum alloy double wire PMIG welding.The dynamic changing of welding current and arc voltage signals can reflect the performance of the welding arc to a certain extent,such as the changing of arc length,and whether there are short circuit spatters,arc breaking and the abnormal droplet transfers etc.In this paper,the welding current and arc voltage signals of aluminum alloy double wore PMIG welding were statistical analyzed,including the probability density analysis of voltages and the UI cyclic graphs of both front and back arcs.And then the qualitative relationship between the voltage probability density curves,the UI cyclic graphs and the stability of the welding process was obtained.Welding is a nonlinear process with strong interference,and the transient welding process parameters are typical nonlinear and non-stationary.Thus,variational mode decomposition(VMD)-a novel time-frequency analysis method is introduced to analyze the welding current signal.The key parameter of VMD algorithm-K was determined by the observing of the signal frequency spectrum.And then,the DC component and harmonic components of pulse current were obtained by VMD method without the modal aliasing phenomenon.Based on that,the standard deviation entropy and singular spectrum entropy were applied to evaluate the stability of welding process.The following test and analysis showed that the method is effective and stable to a certain extent.Because of the effect of the arc thermal inertia on the arc conductivity,the variation of arc voltage in the pulse current rising section and decrease section is obviously different.Based on that,the equivalent resistance of welding arc which is the quotient of instantaneous arc voltage and welding current was proposed to be one of the basic parameters as the electric parameters analysis of welding process evaluation.Then the approximate entropy,sample entropy and fuzzy entropy of the welding current,are voltage and equivalent resistance were compared;the result shows that the approximate entropy of welding current and arc voltage,and the fuzzy entropy of the equivalent resistance can reflect the welding process stability to a certain extent,and the entropy decreases with the increase of the welding stability.In the above indicators,the equivalent resistance fuzzy entropy is more sensitive to the change of the welding process stability.There are complex interactions between two arcs in the aluminum alloy double wire PMIG welding.So,the cross fuzzy entropy which can represent the relative independent between two time series was introduced to evaluate the welding stability.And the cross fuzzy entropy of welding currents,arc voltages and equivalents of front and back arcs were compared.The result shows that the equivalent resistant cross fuzzy entropy is more sensitive to the stability of the welding process.It can be a good evaluation of double wire welding process stability.And the welding process parameter characteristic values which can reflect the difference of welding stability were used as the input vectors of support vector mathine to evaluate the welding stability.After contrast it showed that the average accuracy was up to 90%when use the equivalent resistant cross fuzzy entropy as the input vector.And the average accuracy was up to 93%when use it together with the equivalent resistant fuzzy entropy as the input vector,and the single classification accuracy rate can up to 98.5%.The synergetic acquisition of welding process parameters and welding pool visual image was realize by detecting the pulse current falling edge of the back arc and counting it,then triggering the industrial camera.Then,the geometric parameters of the molten pool were extracted,and the relationship between them and the energy input to the weld was studied.And the singular spectrum entropy,one dimension entropy and two dimension entropy of welding pool image were calculated to distinguish the differences of welding process stability.And the experiment and analysis can cinfirm that the three entropy values of the welding pool image changes obviously with the changing of welding process stability.The cross fuzzy entropy of equivalent resistance,the one dimension image entropy and the singular spectrum entropy were combined to diagnosis the position where the welding process stability changes.Then the welds with surface cleaness changing and sudden blowing lateral wind were designed to test the position diagnosis.The result shows that the proposed method can successfully diagnose the welding process stability mutation location.