Optimization Of Process Parameters For Inverter Type Double-wire Submerged Arc Welding Based On Arc Energy Characteristics

The inverter type double-wire submerged arc welding (SAW) is a nonlinear,multivariate and multiple factors process, the relationship between the weld qualityand process conditions are very complex. Due to the complexity of equipment and thedifficulty of the adjustment of process parameters, it is difficult to accurately graspthe direction and scale of welding parameters adjustment only by experience andwelding test, a slight deviation of production conditions may cause undercuts andhumps, which is a bottleneck of application of the technology in high-speed weldingprocesses. As a result, the study on optimization of process parameters in invertertype double-wire submerged arc welding is of great significance to guarantee thestability of the high-speed submerged arc welding process and the quality of the weld.Based on the analysis of the characteristics of the welding process energy signal,the thesis study the stability feature extraction of welding arc and the assessment ofthe rationality of process parameters based on the local mean decomposition (LMD), anonlinear mapping model between the process parameters of double-wire SAW andthe stability feature of welding arc is established based on the BP neural network, andcombined with the particle swarm optimization (PSO) algorithm, studies onoptimization of process parameters in inverter type double-wire submerged arcwelding is conducted. The main contents are as follows:1. Ethernet communication is applied to the real-time acquisition of arc currentand voltage signals in the double-wire SAW, an arc energy signal acquisition systembased on Ethernet is designed, combined with the MZE-1000/MZ-1250inverter typedouble-wire high speed SAW test platform, an experiment and test platform of singleand double-wire SAW is putting up, and high speed acquisition and storage of arccurrent and voltage signals in single and double-wire high speed SAW process isrealized.2. According to the characteristic of arc energy signals in double-wire SAW, thefeasibility of local mean decomposition applied to analysis of arc energy signals isanalyzed, aiming at the problems such as distortion and decomposition incompleteof the LMD algorithm applied to analysis of arc energy signals, the LMD methodbased on linear interpolation is proposed, and through the energy entropy calculationof product function decomposed from arc energy signals by LMD, the assessment of stability of welding arc and the rationality of process parameters are realized.3. Use the BP neural network, take process parameters of double-wire submergedarc welding as the input of network, take the LMD energy entropy of current signalsof the lead and trace wire as the output of network, a nonlinear mapping modelbetween process parameters of double-wire SAW and the LMD energy entropy featureof arc current signals is established. Training samples are collected by orthogonalexperiments, specific parameters of the model is determined by network training, andthe model is validated through test samples that it can quantitatively reflect thenonlinear relationship between the process parameters and characteristics of arcenergy.4. Take the nonlinear mapping model of process parameters and characteristics ofarc energy as the objective function, take the minimum of the energy entropycharacteristic as the optimization goal, and take the current and voltage of the leadand trace wire, the wire spacing and the welding speed as the optimization variables,a process parameters optimization model of inverter type double-wire SAW isestablished, and utilize the particle swarm intelligence optimization algorithm to do aglobal optimization solution. Through the comparative welding test of the optimizedprocess parameters and the conventional process parameters, validate the optimizedprocess parameters can ensure a good weld formation quality in high-speed welding.