Intelligent Control Strategies And Performance Evaluation Of Integrated Double Wire Arc Welding Power Source

At present, double wire welding power source has many shortcomings, such as large size,complex collaborative control and low intelligence. And the performance is difficult toquantitatively assess, which restricts the realization of intelligent control and intelligentoptimization in double wire power. In response to these issues, the integrated design ideas areproposed to solve collaborative control problems. Further, key and difficult problems inintelligent control strategies of double wire arc welding power source and quantitativeassessment of process performance are in depth research.The design of integrated double wire arc welding power source based on DSP isproposed. The two current control signals are directly given by single control system thatcontrols two limited bipolar soft switching main circuits simultaneously, which solve thecollaborative communication problems. In order to guide component selection and avoiddesign blindness, modeling and simulation is done. Control system model based on signalflow and main circuit model based on components are established. And the connection anddebug between above two models is realized. The overall simulation model is used to predictthe working process of the double wire welding power.For the purpose of the signal noise reduction, digital filtering algorithms, includingdelay filter improved algorithm and adaptive gradient lattice improved algorithm, are studied.The simulation results show that delay filter improved algorithm has good filtering effects andhigh running speed, but it is influenced easily by current parameters. While adaptive gradientlattice improved algorithm has better performance in terms of filtering effect, the running timeand smooth surface.The study found that each wire feed speed of double wire arc welding power source isdetermined not only by their current size, but also by the other current size and a variety offactors. Through correlation analysis, four main influencing factors of wire feed speed arescreened out, and support vector machine predictive control model of the wire feed speed isestablished. In order to obtain better accuracy of the model, the impact of parameters on model learning precision and generalization ability is studied by grid algorithm. Finally, themodel parameters of support vector machines are chosen by particle swarm intelligenceoptimization algorithm. Welding test results show that the optimized model can obtain betterpredictive control.According to integrated design ideas, the hardware and software of the control system isdesigned. Then the signal characteristics of different current pulse are studied, and adaptivefuzzy PID controller is designed. Finally, online tuning of PID parameters is realized.The experiments show that the static and dynamic characteristics of fuzzy PIDcontroller are superior.Process performance is quantitatively assessed from the point of view of electrical signal.The current stability is analyzed by sample entropy algorithm, and the influence of signalparameters on the sample entropy is studied. Double wire quantitative assessment indicatorsand statistical stability evaluation indicators of the electrical signal are designed. Then theindicators are verified in the contrast to sample entropy indicators by the use of crossvalidation ideas.Starting from the arc sound, the stability and splash of the welding process is analyzed.According to the principle that the arc sound energy increases when splash occurs, statisticalquantitative indicators are designed by the use of the cumulative distribution function of theshort term energy and probability distribution. Further, by the spectrogram, arc sound energyvariation is studied, and peak energy curve of time frequency surface is searched. The stability of the peak energy is calculated by sample entropy, and the corresponding index isdesigned for quantitative assessmentIn order to avoid the subjective assessment of weld quality, combined with national standards for weld defect classification, a fuzzy evaluation method is proposed. The cascade weldfuzzy evaluation model is designed and implemented for quantitative assessmentof the weld defect.The multi information fusion thought for quantitative assessment is presented.Regression model for electrical signal, the arc sound and quantitative results of weld quality isestablished by support vector machine. So the quantitative assessment of double wire weldingperformance is achieved. On the self development welding test platform, the performance of self developedintegrated double wire arc welding power source is tested, and the double wire current phaserelationship welding experiment is done. The process performance is assessed quantitatively,too. The results show that the power performance is good, and the assessment results arereliable. On this basis, a new complementary symmetry transition pulse current controlmethod for double wire welding is designed. And the method is analyzed by orthogonal test.Trials have shown that new technology has certain advantages in stability and weld molding.In summary, the integrated double wire welding power source solves the problem ofcomplex collaborative control. The intelligent optimization control strategies, such as digitalfiltering of double wire power input signal, support vector machines prediction of the wirefeed speed and fuzzy PID control, lay the foundation for the further development of the intelligent double wire welding power. The method of multi information fused with Electricalsignals, arc sound, quantitative assessment indicators of weld quality and support vectormachine find a new way for quantitative assessment of the welding performance.