In this work, the forward and reverse mapping of the E-jet-based micro-additive manufacturing process is carried out. A conductive polymer named PEDOT:PSS (poly (3, 4-ethylenedioxythiophene): polystyrene sulfonate) is used to generate patterns on a flexible substrate that can be employed for printed electronics applications. The aim of the study is to demonstrate the use of machine learning techniques for better regulation of E-jet. Artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) are used to model the manufacturing capability of E-jet. Experiments are conducted on the developed experimental setup. The investigated data are used to train and test the machine learning models. Different activation functions and membership functions are used in developing the models. The results are compared to measure the efficacy of the adopted approaches. The comparative study highlighted that the ANFIS model performed better in forward mapping. The model deviation is less than 5%. Furthermore, the bipolar sigmoid ANN-based approach is more suitable for reverse mapping of E-jet. The estimated error for the bipolar sigmoid model is <10%. The present work demonstrated the successful implementation of a machine learning approach for effective offline control of the micro-additive manufacturing process. Practitioners may employ the conducted research work as a template for improving the fabrication performance of E-jet across a wide range of operating scenarios.
Das et al. (Thu,) studied this question.