Dry sliding wear characteristics evaluation and prediction of vacuum casted marble dust (MD) reinforced ZA-27 alloy composites using hybrid improved bat algorithm and ANN

The training process in an artificial neural network (ANN) is regarded as one of the challenging tasks in machine learning due to the complex, non-linear nature and uncertainty involved in determining optimal set of significant governing factors such as number of neurons, weights and biases. In the same context, this paper presents a novel hybridization of an improved bat algorithm (IBA) trained artificial neural network for predicting wear properties of marble dust reinforced ZA-27 alloy composites. The improvement in bat algorithm is performed by introducing a new velocity, position search equation and sugeno inertia weight, that can boost the flexibility and diversity of bat population and provides stabilize effective training of ANN model. In this research, the influence of varying wt. % (0, 2.5, 5, 7.5 and 10 %) of marble dust reinforcement on the wear performance of ZA-27 alloy composites is determined. The prediction of specific wear rate is performed using Taguchi L-25 orthogonal array design and IBA-ANN methodology by performing experimental trials on pin-on- disc tribometer at five levels of each sliding velocity, filler content, normal load, sliding distance and environment temperature. The experimental and predicted values of specific wear rate shows good agreement with an overall accuracy of 96.59 % and 3-d surface plots were established for predicting specific wear rate as a function of wt. % of marble dust and other testing conditions. The results prove effectiveness of IBA trained ANN in overcoming the drawback of local optima stagnation and enhancing convergence speed and can be established as an intelligent tool for prediction of specific wear rate in marble dust reinforced ZA-27 alloy composites.