Data Science and Chemical Kinetics

The constantly increasing pool of experimental datasets in combustion experiments is opening new frontiers in the development of detailed kinetic models: first, their use for the evaluation of kinetic models must be properly automated to avoid the manual validation of the model being the bottleneck of their development. To address such validation, this thesis will develop an ad-hoc procedure for representing combustion chemistry by means of data-driven techniques. Furthermore, the possibility of evaluating the performance of a specific model against a huge amount of data opens the way for the development of new optimized data-driven models through the exploitation of state-of-the-art machine learning algorithms. This will allow to develop extremely reduced kinetic models, thus extending the application of kinetic models for large scale and industrial applications.