Authors: Dheivendran .R, Mostafa Abbas Jabbar
Abstract: Selection of machining parameters in CNC processes has an enormous impact on the resulting surface quality, tool life, and process efficiency. Traditional parameter selection process is mainly driven by operator experience and experimental testing. Such a process takes a lot of time and lacks universality since it can be implemented only in particular situations, taking into account specific material and cutting tool parameters. In this paper, we propose a framework for intelligent machining parameter selection based on integration of machine learning regression techniques with meta-heuristic optimization. The proposed framework uses Random Forest regression model, which is built on the basis of collected machining data. The optimization of machining parameters is performed with the help of Particle Swarm Optimization. The proposed framework is benchmarked against the two baselines, namely response surface methodology and genetic algorithm, in terms of convergence properties, distribution of Pareto front points, and parameter sensitivity rankings. The experiments show that ML-PSO framework explores the space of parameter values more efficiently and finds better surface quality points than traditional techniques.