Gareth Kennedy

data all-rounder and computational scientist


Monash ElectroChemistry Simulator (MECSim)

The Monash Electrochemistry Simulator (MECSim) has software has recently been upgraded to branch out into machine learning. The first step in this direction was to use MECSim to generate sufficient dc cyclic voltammogram data for E, EE and EC mechanisms to be distinguished using a deep neural network (DNN).

This mimics the way an experienced electrochemist will begin their investigation of experimental results by visually inspecting the voltammogram. The next stage would then be to optimize the parameters for the most likely mechanism (e.g. the rate constant for an EC mechanism) and obtain the best fit between the simulated waveform and the experimental results. The aim of the machine learning pipeline being built using MECSim is to automate and streamline this process. A proof of concept for the first phase has been completed (see publications) which allows these mechanisms to be identified automatically within 5 ms. The full pipeline is currently being built in collaboration with the Electrochemistry Group, School of Chemistry, Monash University and the Mathematical Institute, University of Oxford.

A paper outlining our approach has just been submitted (Nov 2020). Slides from an overview talk on "Applying Machine Learning to Experimental Results from Analytical Chemistry" presented as part of the ARC Centre of Excellence for Electromaterials Science (ACES) webinar series can be found here.

Back to top