An Application of Computational Intelligence Techniques to Predict Biometal Deposition Characteristics in Metal Additive Manufacturing

In metal additive manufacturing process, deposition characteristics are very important to determine the quality of the deposition and the process efficiency. To achieve, this an adaptive neural-fuzzy-based computational intelligence (CI) technique is used to model the layer width and layer height in laser-based direct energy deposition technique as this is one of the most promising metal additive manufacturing methodologies. In this present study, neural network-based architecture was applied to develop an optimal neural-fuzzy system for modeling and predicting the width and height of the deposited layer for three controllable process variables such as feed rate, power, and scanning velocity of the laser. Hybridization of steepest descent and least squares method was used as a learning method in the proposed adaptive neural-fuzzy predictive system. To do the comparative study of prediction accuracy of biometal deposition characteristics, trapezoidal and Gaussian membership function were considered after careful study. The predicted deposition characteristics feature values predicted from proposed adaptive neural-fuzzy model have been compared with the real data. This comparative study indicates that the proposed methodology can design optimal data base and rule base of the fuzzy system for predicting the deposition characteristics in metal additive manufacturing process.