Study On Numerical Simulation And Optimization Design Of Laser Bending Of Titanium Alloy Sheets

Laser bending is an efficient flexible manufacturing technology,which is realized by introducing thermal stress into the surface of a workpiece without melting. Forming is achieved by scanning with a high power laser beam over the surface of the material. This technology is especially capable of shaping hard and brittle or other materials with great rigidity, such as ceramics and titanium alloys.TA15 alloy is such a new near-alpha titanium alloy which not only offers the moderate strength and good thermal stability, but has excellent heat resistance and weldability like alpha titanium alloys and technical plasticity close to alpha-beta titanium alloys. Because of its excellent properties, it has been found an increasingly wide application in aerospace industry, but its cold formability is bad and rebound is very high. So titanium alloy is commonly hot formed. The cycle of the laser bending production is shorter and the cost is lower than that of conventional hot-forming technology. Therefore, it is realistically significant to apply laser bending to titanium alloys, and the unique dominance of this kind of new hot-forming technology could be made full use of.In this paper, a three-dimensional thermal-mechanical coupled finite element analysis model of TA15 alloy sheets of multiple-irradiation laser bending process was developed by the software MSC. Marc. The variations of temperature field, deformation field and stress field were numerically calculated. And the effects of different process parameters, geometric parameters and scanning traces on laser bending process were studied. And then different parameters were investigated by the orthogonal test and a combination of optimum parameters were obtained under given conditions.Then a BP network model and a GA-BP network model were individually put forward for predicting the laser bending angles in order to comprehensively utilize global optimizing abilities of Genetic Algorithm and local searching abilities of BP Neural Network. In this paper, both of the models were individually applied to predict the laser bending angles by selecting the main influencing factors as the input variables and the bending angles as the output variable. Based on MATLAB software, the prediction function of the two network models were realized.By comparing GA-BP network model with BP network model, the results show that applying the two networks to predict laser bending angles is feasible and efficient, the speed of operation of GA-BP network is much faster and the precision of prediction of GA-BP network is much higher.