Article Preview
Top1. Introduction
Optimization is a very important issue in mechanical industry. Specially, in machining processes, where different aspects must be considered, selecting the most proper cutting conditions plays a key role for obtaining efficient and competitive products.
From the first published cutting processes optimization work (Taylor, 1907), to the most current literature (Rao & Kalyankar, 2014), a big amount of research has been expended on this topic. Nevertheless, a general robust solution, which can be applied to any cutting process, has not been obtained (Umer, Qudeiri, Hussein, Khan, & Al-Ahmari, 2014). This is caused by two main reasons. On one hand, the high complexity of the phenomena involved on the cutting processes makes it very hard to obtain reliable accurate models for describing the behaviour of the involved variables. Most of the optimization approaches use empirical models which cannot be applied outside the data intervals used for fitting these models. Moreover, obtaining experimental data from cutting processes is expensive and time consuming.
Other approaches have been proposed for modelling the cutting processes, such as finite element method (Abouridouane, Klocke, & Döbbeler, 2016; Amrita, Surjya, & Arun, 2011; Bartarya & Choudhury, 2011; Markopoulos, Kantzavelos, Galanis, & Manolakos, 2011; Umer, 2016; Umer et al., 2014; Weng, Zhuang, Chen, Guo, & Ding, 2017) and mechanistic approaches (Abouridouane et al., 2016; Bai, Sun, Roy, & Silberschmidt, 2017; Baohai, Di, Xiaodong, Dinghua, & Kai, 2016; D’Acuntoa, Le Cozb, Moufkib, & Dudzinski, 2017; Fu, Chen, Mao, & Xiong, 2017; Gao, Sun, & Leopold, 2018; Vinogradov, 2014; Weng, Zhuang, Zhu, Guo, & Ding, 2018; Zhang & Guo, 2015). Both offer reasonable accuracy with few experimental data, but requires a lot of time for computing the outcomes (specially, the finite element method).
On the other hand, due to the complex nature of these models, the objectives functions and constraints do not fulfil the conditions of continuity, differentiability and unimodality, required for most of the optimization tools (Quiza, López-Armas, & Davim, 2012). Therefore, heuristics techniques have been applied for solving these problems. They include genetic algorithms (Batish, Bhattacharya, Kaur, & Cheema, 2014; Ganesan & Mohankumar, 2013; Kübler, Böhner, & Steinhilper, 2015), simulated annealing (Baseri, 2011; Wang, Wong, Rahman, & Sun, 2006), particle swarm optimization (Marko et al., 2014) and ant colony optimization (Vijayakumar, Prabhaharan, Asokan, & Saravanan, 2003), but all of them rely on the accuracy and reliability of the underlying models. Furthermore, the use of a single objective model is not enough for depicting the complex nature of the cutting processes (Kovačević, Madić, Radovanović, & Rančić, 2014; Saha & Majumder, 2016). Neither the a priori multi-objective optimization, which combines the different goals into a single one, can solve this problem (Quiza, Beruvides, & Davim, 2014).
The present paper proposes a hybrid approach for modelling and optimizing oblique cutting processes, which combines analytic, empirical and heuristic techniques. This method avoids the use of large and expensive amount of experimental work and allows carrying out the optimization with accuracy and flexibility.