Analysis of laser powder deposition parameters: ANFIS modeling and ICA optimization

Abstract

The newly developed laser powder deposition technique requires the optimization of various coating parameters. The optimized parameters will lead to a greater coating performance. We present an adaptive neuro-fuzzy inference system, ANFIS, to model the response of deposition parameters based on experimental data of Laser powder deposition of Fe-based alloy on ASTM 36 mild steel substrate. A series of systematic experiments has been conducted using central composite design (CCD) taking into account the response of laser power (LP), powder feeding rate (PF), carrier-gas flow rate (CG) and stand-off distance (SD) as processing parameters and catchment efficiency (CE), clad height (CH) and clad width (CW). Moreover, in order to simultaneously maximize these parameters, the ANFIS models of responses were associated with imperialist competitive algorithm (ICA). Results indicated that ANFIS structures with 2 2 2 2 Triangular membership function ensure lowest values of prediction error for catchment efficiency and clad width, while for clad height 2 2 2 2 structure with Generalized-bell, the membership function has highest prediction accuracy. Also, ICA optimization show that setting LP=3.199kW, PF=43.237g/min, CF=19.583 SCFH, and SD=8.632 mm leads to maximum responses in CE=0.5417, CH=1.3470mm and CW=12.3048mm. The modeling results are in agreement with the experimental data.