Machinability study on FRP composites – A neural network analysis

Abstract

Cutting force is one among the basic factors, which measures the machinability of any material. The determination of cutting forces by experimental means is more common as there is no general mathematical model is available to predict the cutting forces developed as a function of machining and material parameter. However conducting experiment to measure these forces every time is highly expensive and time-consuming. In some cases particularly in machining composites, availability of specimens with the required specifications is extremely difficult. In this work, a multi-layer perception feed forward neural network is constructed to evaluate and compare the cutting forces developed during the machining of glass/epoxy, graphite/epoxy and kevlar/epoxy composites. The fibre orientation, volume fraction and depth of cut were chosen as input parameters for this purpose. The cutting force values evaluated by finite element analysis have been used for training the network. The neural network outputs are compared with the desired output values. Maximum error reduction is possible. Further, the comparison of the neural network output results with the results obtained from experiments and empirical relationship has shown good agreement. Results revealed that kevlar/epoxy has shown good machinable characteristic.