Data-driven framework of ECM: A machine learning model for profile prediction

Electrochemical machining (ECM) with pulsed power supply can decrease the stray current corrosion, and thus, provides better surface quality as well as machining accuracy. However, the non-linear nature of electrochemical reactions makes it challenging to predict the final profile of the machined feature through the typical multi-physics based finite element (FEM) model. It is difficult to simulate the ECM process across wide range of processing parameters in the state-of-the-art commercial FE softwares due to several issues related to timestepping, convergence, excessive mesh distortion during moving mesh, model insensitivity to small change in parameters etc. In this article, we present a framework of data-driven ECM model, which is based on the supervised machine learning approach, for predicting the final profile of the machined feature in ECM process. After training this data-driven ECM model with different levels of pulsed peak voltage and electrolyte flow rate, predictions and experimental validations are conducted even outside the parameter window to demonstrate the performance of generalization and applicability to a wide range of machining parameters of our data-driven ECM model.

ISBN:978-9-08289-313-7

Jaar van publicatie:2021

Toegankelijkheid:Embargoed