Use of Machine Learning Methods for Predicting Amount of Bioethanol Obtained from Lignocellulosic Biomass with the Use of Ionic Liquids for Pretreatment

The study objective was to model and predict the bioethanol production process from lignocellulosic biomass based on an example of empirical results. Two types algorithms were used in machine learning: artificial neural network (ANN) random forest algorithm (RF). Data for included results studying with use ionic liquids (ILs) different enzymatic preparations following types: buckwheat straw four wastelands, including a mixture various plants: stems giant miscanthus, common nettle, goldenrod, broom, fireweed, hay (a mix grasses). input variables consisted (imidazolium ammonium), preparations, enzyme doses, time temperature pretreatment, type yeast alcoholic fermentation. output value concentration. multilayer perceptron (MLP) networks. created; training dataset comprised 120 vectors (14 elements Model 1 11 2). Assessment optimum carried out using same division experimental points (two datasets, containing 2/3 1/3 testing) criteria models. mugwort hemp validation. In both models, coefficient determination networks <0.9, while RF it oscillated around 0.95. Considering fairly large spread coefficient, two hybrid models generated. approach creating describing present resulted increase fit R2 = 0.961. can be initial classification plants without necessity perform lengthy expensive research related IL-based pretreatment further hydrolysis; only their composition are needed.