A Hybrid Approach for Modelling and Optimization of Laser Cladding Process

Abstract

To develop the hard surfaces of superalloys usually, laser surface modification is employed. Out of several laser surface modification techniques, the laser cladding process allows obtaining sound surface clads. Laser Cladding being a complex process depends on many input parameters. Knowledge of these input parameters during laser cladding is essential for the development of good clad. Artificial intelligence has proven to be a better tool for modeling processes having complex and non-linear behavior. A hybrid approach of any two individual methodologies may even perform better. In this paper, a coupled methodology of Artificial Neural Network (ANN) and Particle Swarm Optimization (PSO) for modeling and optimization of process parameters (laser power, scan velocity, and powder feed rate) for quality characteristics (aspect ratio) in laser claddingInconel-738 is proposed. First, an ANN model is trained, tested, and developed for aspect ratio. Then, the PSO technique is used for optimization utilizing a trained ANN model. The developed hybrid model shows the minor error of 5.13% of mean square error and 8.68% error in predicting aspect ratio