Multi-objective optimization of multi-axis ball-end milling Inconel 718 via grey relational analysis coupled with RBF neural network and PSO algorithm

Multi-axis ball-end milling is the most commonly used operation in machining aerospace engine parts. Because of multi-output characteristic, the process improvement often requires multi-objective optimization. Recently, the grey relational analysis (GRA) has been more and more widely used in engineering manufacture with multiple responses. But, the original GRA method only suits for the optimization problem in discrete space. This paper proposes an integrated multi-objective optimization method with GRA, radial basis function (RBF) neural network, and particle swarm optimization (PSO) algorithm. Compared with the original GRA, it expands the optimal solution space to continuous space. This approach is subsequently applied to the multi-objective optimization of multi-axis ball-end milling Ni-based superalloy Inconel 718. The purpose is to simultaneously obtain minimum surface roughness and maximum compressive residuals tress by optimizing the inclination angle, cutting speed, and feed. A hybrid experiment scheme with single factor design and orthogonal array is utilized to generate the sample data set. The multi-response optimization problem is successfully converted into the single objective optimization of grey relational grade (GRG). Then, the RBF neural network is employed to establish the mapping relation between the GRG and the process parameters. And its adequacy is proved by five test experiments with a low prediction error of 6.86%. Finally, the PSO algorithm is adopted to optimize the process parameters. Verification experiments show that a higher improvement of the GRG is obtained with the proposed method (62.87%) than that of the original GRA (50.00%). The developed approach is proved to be feasible and can be generalized for other multi-objective optimization problem in manufacturing industry.