An Adaptive Neural-Fuzzy Approach for Modeling of Droplet Frequency in E-Jet-Based Micro-additive Manufacturing

Electrohydrodynamic inkjet printing-based micro-additive manufacturing technology is a nano-manufacturing process using fluid jet printing, influenced by an electric field through nano-scale nozzles. This method helps to achieve higher resolution and control which is way better than that of traditional processes. This paper focuses on a fuzzy logic-based neural networking system, also known as ANFIS which has been used to predict and model the output parameter, i.e., droplet frequency for three controllable processes like nozzle-substrate gap, ink flow rate, and applied voltage. Since it is difficult to maintain constant operating conditions, the jet frequency and droplet diameter fluctuate. In order to stabilize this fluctuating jet frequency, a second-order non-linear regression equation has been implemented. In the proposed adaptive neuro-fuzzy predictive system, hybridization and backpropagation were used to predict the values. To do a comparative study in the accuracy of prediction of the characteristics of the deposited droplet, triangular, Gaussian and bell-shaped membership functions have been considered after a careful study. The comparative study indicates that, for predicting the droplet frequency, the optimal data base and the rule base of the expert system can be designed by the proposed methodology.