基于NSGAⅡ算法和BP神经网络的轮辋焊接工艺参数优化OA

The propensity for cracking at the weld seam during the flaring process after flash butt welding of rims was investigated from two perspectives:the metallographic microstructures of the base metal,weld zone,and heat-affected zone,and the fracture morphology.Based on the analysis,recommendations for optimizing welding process parameters were proposed.To improve the quality of flash butt welded joints of Q355D rim steel,an experimental study was conducted.A BP-NSGAⅡ algorithm was employed to optimize four input process parameters,namely burn-off length,burn-off speed,upset distance,and energized upset time,while elongation and impact toughness were selected as output indicators.The results indicate that the trained BP neural network satisfies the required prediction accuracy.By integrating it with the NSGAⅡ algorithm and performing multiple iterations,an optimal combination of process parameters was obtained.The optimized parameters were determined as follows:a burn-off length of 18 mm,a burn-off speed of 1.1 mm/s,an upset distance of 10.0 mm,and an energized upset time of 0.35 s.Validation through tensile and impact tests showed that the measured elongation reached 25.10%,and the impact absorbed energy was 22.6 J.Application of these optimized welding parameters in production and subsequent tracking demonstrated that the actual cracking rate of rims was effectively controlled below 15%.