In this paper laser welding AA5182 of aluminum alloy with AA5356 filler wire were performed with respect to laser power, welding speed, and wire feed rate. The experiments showed that the tensile strength of the weld was higher than that of the base material under sufficient heat input conditions. A genetic algorithm was used to optimize process parameters which were the laser power, welding speed, and wire feed rate. To do that, a fitness function was formulated, taking into account weldability and productivity. A factor for the weldabilty used tensile strength estimation model which was made by neural network, and as the productivity, welding speed, and wire feed rate were used. Weld monitoring system for aluminum laser welding with filler wire was constructed through the optical sensors to measure the plasma light intensity. Relationship between monitoring signal and plasma and keyhole behavior according to welding condition was analyzed and it was found that sensor signal could express the information for weld quality. Weld quality estimation algorithm was formulated fuzzy multi feature pattern recognition algorithm using the monitoring signals. Quality prediction system was also developed to apply this algorithm to production line.
