Modelling and optimisation of machining parameters for composite pipes using artificial neural network and genetic algorithm

Abstract

An integrated approach to determine the influence of cutting parameters on carbon fiber reinforced polymer (CFRP) pipe using Taguchi’s Design of experiments (DOE), artificial neural network (ANN) and optimisation with genetic algorithm (GA) is followed when carrying out turning operation on a bidirectional carbon fabric reinforced bisphenol pipe using a carbide tool at different set of input parameters like feed, speed and depth of cut. Due to the anisotropic nature of composites and inhomogeneity, optimisation of machining parameters for composites is very difficult and largely dependent on wide variety of combinations among various factors such as fiber properties, orientation, resin properties, curing conditions, etc. For successful application of these composites in different fields, knowledge on machining the composites is necessary. Taguchi’s orthogonal array is used for DOE and the output under consideration is forces acting on cutting tool. Genetic algorithm is used to optimize the machining parameters to yield minimum cutting force. ANN with back propagation neural network algorithm has been adopted to model the machining operation. The results were verified using confirmation tests which indicate that the experimental values and observed values are almost equal with minimum error and ANN with GA could be successfully implemented for modelling and optimizing the machining parameters for bidirectional CFRP composite pipes.