Article Preview
TopIntroduction
Optimization is a basic tool in all areas of applied mathematics, process planning, engineering, medicine, economics and other sciences. The selection of optimum process parameters plays a significant role to ensure quality of product, to reduce cost and to increase productivity. Demand of the optimum decision making is rising continuously due to increasing complexity in the problems and requirement of precise decision. Although in a single objective optimization problem the solution can be straight forward, multi-objective ones provide set of solutions from which the choice is to be made. Problems with multi objectives/criteria are generally known as multiple criteria optimization or multiple criteria decision making (MCDM) problems. The area of multiple criteria decision making has received enormous attention in the recent years, primarily due to rapid development in computer technology, which includes development and availability of user-friendly software, high-speed and parallel processors etc. Depending upon the characteristics of the problems, these can be sub divided into two types: multi-attribute decision analysis and multi-objective optimization. In multi-attribute decision analysis, the set of the feasible alternatives is discrete, predetermined and finite. In multi-objective optimization problems, the feasible alternatives are not identified in advance among the number of known solutions. These problems can be called continuous and one has to generate the alternatives before they can be evaluated (Kalsi et al., 2014; Raykar et al., 2015).
Till now, various optimization techniques have applied by the researchers in machining. Accepting the best solution after comparing a few solutions is the indirect way of attaining optimal solution in many industrial problems. However, there is no way of guaranteeing an optimal solution with this basic approach. Aggarwal & Singh (2005) in their review discussed different conventional techniques employed for optimization in machining, which includes geometric programming, linear programming, goal programming, sequential unconstrained minimization technique, dynamic programming. The latest are fuzzy logic, artificial neural network technique, scatter search technique, genetic algorithm, Taguchi technique, response surface methodology and their combinations etc.
Lin et al. (2001) used convergence network and constructed a simulation model for cutting forces and surface roughness, considering cutting speed, feed rate and depth of cut as input parameters. A regression analysis was adopted in the study to develop a second prediction model to verify the accuracy of the network. Comparison of these confirmed that prediction model developed by the network was more accurate than that of the regression analysis. Genetic algorithm (GA) technique aims at selection of the optimal values of the input parameters to get acceptable output responses in machining (Shunmugam et al., 2000; Davim & Antonio, 2001).
Neural network basically uses models that simulate the working model of the biological neurons. The modern usage of the term often refers to artificial neural networks (ANN), which consists of artificial neurons or nodes. The numbers of neurons are adjusted or trained to minimize the error between the predicted inputs and outputs for a specific target output. Once trained, the neural networks can be used for prediction of optimal solution (Dixit & Dixit, 2008; Joshi & Pande, 2009).
Fuzzy logic theory is another method of interest used by many researchers for optimization by considering the uncertainty in processes (Lin et al., 2000; Saaty & Tran, 2010; Uthayakumar & Valliathal, 2011). Jenab et al. (2012).