Predicting detonation cell size of biogas–oxygen mixtures using machine learning models

Abstract

Detonation cell size is a very important parameter describing the detonation process, used both for explosion safety analysis and for the design of detonation combustion chambers. Typically it has been studied either experimentally or by CFD simulations; both options are costly in terms of money and time. However, progress in the machine learning (ML) methods opened a third way of obtaining cell size. When trained properly, such models are capable of giving rapid, accurate predictions. Utilization of machine learning in the combustion field is gaining more attention from the research community. In this study, the process of training, testing, and evaluation of three different machine learning models for predicting biogas–oxygen mixture detonation cell size is presented. The models include: linear regression (LR), support vector regression (SVR), and neural network (NN). The dataset used for training and testing comes from the experimental studies conducted previously by the authors. It was shown that all the models give very good results with support vector regression proving to be the best.