Prediction of Rate of Chemical Reaction using Computational Intelligence

Abstract

Rate of chemical reaction is one of the most important thing in chemical kinetics. Rate of reaction determines how fast or slow a reaction is taking place. Traditionally, Arrhenius equation is used to know rate of chemical reaction at different temperature and concentration. However, finding values of parameters of Arrhenius equation for a reaction can be costly as well as time consuming. The work presented in this thesis mainly focuses on predicting value of rate of reaction using machine learning techniques. The objective of this thesis is to find optimum parameters for rate of reaction from easily measurable features of chemical reaction, such as temperature, pressure, density and concentration of reactants involved in a chemical reaction. First of all, we collected data by simulating chemical reaction using Cantera. We choose 5 different reaction and simulated them in Cantera. For simulating these 5 reaction, we created an input file which have data about these 5 chemical reaction and then this input file was given to Cantera. Then we varied temperature, ressure of environment and changed the amount of reactants being used for chemical reaction then we notice corresponding rate of reaction. We applied various machine learning models on data obtained from simulation to predict rate of reaction and compared their performances with each other to find the best machine learning model. To check the robustness of best model, we used k-fold cross validation.