An effective approach of adaptive neuro-fuzzy inference system-integrated teaching learning-based optimization for use in machining optimization of S45C CNC turning

This paper proposes an effective integration of the Taguchi method (TM), Adaptive neuro-fuzzy inference system (ANFIS) and Teaching learning-based optimization (TLBO) for CNC turning optimization of S45C carbon steel. The TM plays two main roles: it reduces the number of experiments and identifies the most appropriate membership functions (MFs) and suitable learning procedure for the ANFIS. To determine the suitable ANFIS structure, we optimize the root mean squared error, a performance criterion of the ANFIS. Then, taking the established ANFIS structure, we form the virtual mathematical relations between the geometric parameters and the roughness surfaces. The results found that the triangular-shaped MFs and π-shaped MFs are the best for the R a and R z roughness surfaces, respectively. The optimal parameters for ANFIS structure of R a are found in terms of the number of input MFs of 3, the trimf MFs, hybrid learning method, and linear output MFs. The optimal parameters for ANFIS structure of R z are determined at the number of input MFs of 3, the pimf MFs, hybrid learning method, and linear output MFs. Based on the improved ANFIS establishments and optimal parameters of TLBO, the TLBO-based ANFIS is used to optimize the design parameters of the turning. We apply analysis of variance to determine the significant contribution of each factor. The results show a relative decrease in the roughness surfaces compared to those predicted by other algorithms. Therefore, the proposed optimization approach is a robust and effective tool for engineering applications.