2022 Volume 63 Issue 742 Pages 157-162
Peen forming has been widely applied to aircraft wings. In conventional peen forming using spherical shots, axisymmetric plastic strain causes spherical deformation. A new peen forming method, vibration peening, using a rectangular solid pin was developed. In this method, the curvature radius Rx in the x-direction (short side of the pin) was smaller than the curvature radius Rz in the z-direction (long side of the pin). In vibration peening, the pin is projected onto the specimen using a hammer with a reciprocating motion. For a plate of aluminum alloy A7075-T6, the effects of the pin movement (x-direction speed vx and z-direction travel pitch Pz ) and the parallel length L of the pin tip on 1/Rx and 1/Rz were investigated. To predict 1/Rx and 1/Rz from forming conditions, machine learning regression analysis (support vector regression) was performed using the experimental results of vx , Pz , L, 1/Rx and 1/Rz as training data. Regression analysis verified that Rx and Rz were approximately predictable under the unlearned condition. Therefore, regression analysis can be used to select the appropriate forming conditions (vx , Pz , L ) for achieving the target curvature (1/Rx and 1/Rz ).
