Achieving maximum dimensional accuracy and surface quality at the shortest possible time in single-point incremental forming via multi-objective optimization

Abstract

Single-point incremental forming is a novel and flexible method for producing three-dimensional parts from metal sheets. Although single-point incremental forming is a suitable method for rapid prototyping of sheet metal components, there are limitations and challenges facing the commercialization of this process. Dimensional accuracy, surface quality, and production time are of vital importance in any manufacturing process. The present study is aimed at selecting proper forming parameters to produce sheet metal parts which possess dimensional accuracy and good surface quality at the shortest time. Four parameters (i.e. tool diameter, tool step depth, sheet thickness, and feed rate) are chosen as design variables. These parameters are used for the modeling of the process using Group Method of Data Handing(GMDH) artificial neural networks. The data necessary for establishing empirical models are obtained from single-point incremental forming experiments carried out on a computer numerical control milling machine using central composite design. After the evaluation of the model accuracy, single- and multi-objective optimization are performed via genetic algorithm. The performance of the design variables of a tradeoff point corresponding to one of the experiments shows the efficiency and accuracy of the models and the optimization process. Considering the priorities of objective functions, a designer will be able to set proper process parameters.

Modeling and Analysis of Single Point Incremental Forming Force with Static Pressure Support and Ultrasonic Vibration

“…It was found that the tool head diameter had the most significant effect on the geometric accuracy. Abolfazl Taherkhani et al [11] used the group method of data handling (GMDH), an artificial neural network to model the relationships between tool head diameter, layer spacing, sheet thickness, feed speed, axial accuracy, and surface quality. The model was validated by experiments.…”

Sensitivity Analysis and Optimization of the surface Roughness in the Incremental Forming of mild Steel Sheets

“…L 16 Taguchi orthogonal array is adopted to investigate how the four input variables affect the SR of the mild steel sheets. The control factor values for each SPIF experiment are shown in Table 3.…”

“…The results illustrated that the most important parameter that influenced the surface roughness was the step down. Taherkhani et al [16], using the Al3105 sheets, performed four process parameters to model the Group Method of Data Handling (GMDH) as a sub model of artificial neural networks.…”

“…Input parameters and their levels in the Taguchi design of experiments (4 factors in 4 levels) L16 Taguchi orthogonal array designed for the roughness measurement on the mild steel sheets Table of responses for means of R a in the SPIF process of mild steel sheets Analysis of variance of R a of the mild steel sheets in the SPIF experiments of forming the elongated cone Table of Responses for the signal-to-noise ratios in the-smaller-the-better case…”

Mechanism and modeling of the multi-stage bending process for large-scale forgings

“…As for the die design and optimization, the metamodeling and optimization methods, which are also called as surrogate-assisted optimization, have been extensively applied in laboratories and industries due to the high efficiency and feasibility in dealing with the strong nonlinear, high-dimensional and large-scale metal forming problems. 4–6 Zhao performed a multi-objective optimization for porthole extrusion die based on Kriging metamodeling and Pareto-based genetic algorithm (GA) to decrease the die stress and mandrel deflection and even the velocity distribution in the cross section of the profile. 7 Sun 8 investigated the effects of geometry of second-step welding chamber of the extrusion die on the standard deviation of the velocity field in bearing exit and obtained the optimum parameter set by combining Box–Behnken experimental design with response surface method.…”