Modeling and optimization of ultrasonic metal welding on dissimilar sheets using fuzzy based genetic algorithm approach

Ultrasonic welding has been used in the market over the past twenty years and serving to the manufacturing industries like aviation, medical, microelectronics and many more due to various hurdles faced by conventional fusion welding process. It takes very short time (less than one second) to weld materials, thus it can be used for mass production. But many times, the problems faced by industries due to this process are the poor weld quality and strength of the joints. In fact, the quality and success of the welding depend upon its control parameters. In this present study, the control parameters like vibration amplitude, weld pressure and weld time are considered for the welding of dissimilar metals like aluminum (AA1100) and brass (UNS C27000) sheet of 0.3 mm thickness. Experiments are conducted according to the full factorial design with four replications to obtain the responses like tensile shear stress, T-peel stress and weld area. All these data are utilized to develop a non-linear second order regression model between the responses and predictors. As the quality is an important issue in these manufacturing industries, the optimal combinations of these process parameters are found out by using fuzzy logic approach and genetic algorithm (GA) approach. During experiments, the temperature measurement of the weld zone has also been performed to study its effect on different quality characteristics. From the confirmatory test, it has been observed that, the fuzzy logic yields better output results than GA. A variety of weld quality levels, such as “under weld”, “good weld” and “over weld” have also been defined by performing micro structural analysis.