Transfer Learning Combined with High-Throughput Experimentation Framework for Integrated Biorefinery

Abstract

The need of the hour is to maintain a dynamic equilibrium between man and universe for sustainable life. There is an urgent necessity in providing humanity with new materials and clean energy in a sustainable way. In other words, it is the integration of environment, energy, equity and economy which is known as Green Technology. This will require more efficient and entirely different use of natural resources such as abundantly available lignocellulose biomass waste. Novel synthesis routes for integrated bio refinery plants will have to be developed in which waste streams can be converted into essential fuel and chemical streams to fulfill the needs of the society. The development of new manufacturing processes critically depends on the rational design and development of new catalysts. Catalytic materials accelerate and facilitate the conversion of raw materials into products at milder process conditions and with reduced energy consumptions. However, traditionally, catalyst development is still carried out using trial-and-error methods, few empirical models (Nolan et al. in Nature Catalysis, 2018), which are slow, undirected and unreliable. The perspective is toward exploiting waste biomass resources, to design and develop a rational catalyst and efficient transfer learning approach for integrated biorefinery applications. Combine hybrid models, machine learning algorithms and high-throughput experimentation are studied in the past and provide the enabling factors for new material and energy streams to have ‘plenty for all and perennially’ (https://www.ncl.ac.uk/).