A Study on Different Activation Functions for Neural Network-Based Wear Loss Prediction

Inspired by the biological nervous system, an artificial neural network (ANN) is a fascinating mathematical tool, which can be used to simulate a wide variety of complex scientific and engineering problems. The potential of using neural network in prediction of wear loss quantities of A390 aluminium alloy and the suitability of different activation functions like binary sigmoid, identity and radial bias functions has been studied in the present work. The material is subjected to dry sliding wear test using pin-on-disc apparatus at room conditions. Effects of load, sliding speed and time have been investigated by using artificial neural networks using different activations. The experimental results were trained in an ANNs program and the results were compared with experimental values. It is observed that the experimental results coincided with ANNs results and binary sigmoid function proves to be a better activation function for the wear loss prediction for A390 aluminium alloy.