In the existing automobile manufacturing process, the welding connection between each part has been generally made by the metal inert gas (MIG) and tungsten inert gas (TIG) welding method. These welding methods are fusion welding, and heat input is very high, and it causes a phase change of the base material because they melt the contact surface of the parts. Therefore, the thermal deformation of the product is large and the residual stress in the area near the weld is high, resulting in a decrease in mechanical strength, and quality problems such as pores and cracks frequently occur. In this study, friction stir welding (FSW) was applied to the welding process about some parts of the newly developed next-generation electric vehicle manufacturing process to compensate for the problems of the existing fusion welding method. The part where FSW is applied is the four fillet joint areas where the side frame and the bottom frame constituting the battery frame are contacted. In the case of the side frame, the cross-section is complicated to secure the rigidity of the vehicle, and it is made of thin aluminum extrusion material for weight reduction. In FSW, stir occurs easily only when the material is softened by frictional heat caused by rotation of the tool, and the overall welding proceeds smoothly. Frictional heat is generated on the surface where the rotating tool and the material come into contact but In the case of the Fillet FSW, the contact area is reduced by more than 70% compared to the butt FSW due to the shape of the tool. So, in order to apply Fillet FSW to an actual structure, it is necessary to derive the temperature distribution through basic experiments and numerical analysis and additionally predict the behavior of the entire structure. First, the fillet FSW experiment was conducted in units of the specimen to derive the optimal FSW welding process and measured the temperature. In addition, through numerical analysis using the Reynolds model, the temperature of the welding center area was derived and compared with the measurement result to reduce the error. Subsequently, the deformation and residual stress were calculated by performing thermal elasto-plastic analysis on the 3D design model of the entire battery frame structure, and the deformation was measured by performing Fillet FSW on the actual battery frame structure. Finally, the analysis result for the deformation was confirmed to have very little error compared to the measured value. The numerical analysis method proposed in this study is expected to improve the design accuracy of the Fillet FSW structure, thereby reducing the test period and manufacturing process cost and increasing productivity.