Artificial neural network based modeling for the prediction of yield and surface area of activated carbon from biomass

Abstract

Activated carbon (AC) is an adsorbent material with broad industrial applications. Understanding and predicting the yield and quality of AC produced from different feedstock is critical for biomass screening and process design. In this study, multi‐layer feedforward artificial neural network (ANN) models were developed to predict the total yield and surface area of AC produced from various biomass feedstock using pyrolysis and steam activation. In total, 168 data samples identified from experiments in literature were used to train, validate, and test the ANN models. The trained ANN models showed high accuracy (R2 > 0.9) and demonstrated good alignment with the independent experimental data. The impacts of using datasets based on different biomass characterization methods (i.e., ultimate analysis and proximate analysis) were evaluated and compared. Finally, a contribution analysis was conducted to understand the impact of different process factors on AC yield and surface area. © 2019 Society of Chemical Industry and John Wiley & Sons, Ltd

Impacts of uncertain feedstock quality on the economic feasibility of fast pyrolysis biorefineries with blended feedstocks and decentralized preprocessing sites in the Southeastern United States

“…In total, 64 samples of ASPEN Plus simulations using different combinations of biomass feedstock composition (including carbon content, MC, and ash content) (see Table S7) were obtained as the training dataset for ANN. ANN is a machine learning approach that has been used by previous studies to predict yields and product compositions of pyrolysis (Arumugasamy & Selvarajoo, 2015; Conesa et al, 2004; Hough et al, 2017; Karaci et al, 2016; Liao et al, 2019; Sun et al, 2016; Sunphorka et al, 2017). The structure of ANN model in this study consists of input layer (containing three input variables), hidden layer (containing 10 neurons), and output layer (Sunphorka et al., 2017).…”

“…The R 2 value of gasoline yield, diesel yield, natural gas consumption, and net power output is .98, .96, .98, and .96, respectively. To avoid the potential overfitting issues, the extra validation was conducted following the procedure in the study (Liao et al., 2019). Additional 17 runs (see Table S8) were conducted using the datasets generated by ASPEN Plus simulation models but not included in training datasets of ANN.…”

Fuel Classification based on Flame Characteristics using a Time Series Analysis with Fuzzy Support Vector Machine Algorithm

“…C is the default parameter of the cost control, and m is the Fuzzy index (m = 2) in Equations (20) and 21, respectively. S i (new membership value) is calculated for data using the formula given in Equation (21). These values of S i are given in the constraints of SVM.…”

“…Step 5: Use FSVM algorithm for classification of fuel types //using Eq. (13)(14)(15)(16)(17)(18)(19)(20)(21) 3 | EXPERIMENTAL RESULTS…”

“…Algorithms such as Cellular Neural Network, 15 Neural Network Modeling, 16,17 Euclidean Distance Classifier, Principle Value Decomposition, 10 Self-Organizing-Feature-Maps, Multivariate Analysis, Fuzzy Cmeans Clustering, Fuzzy Fusion control, Hidden Markovian model, and Maximum Posterior Marginal 18 were used to train and classify the images based on fuels and flame identification. Linear and Logistic regression model, Multinomial-expression, Constructed inference algorithm, Principle Component Analysis, Genetic Algorithms, Artificial Neural Networks, [19][20][21] Fuzzy Logic, Clustering concepts, Naïve-Bayes, Stochastic models, Support Vector Machine (SVM) algorithm, and other Machine Learning algorithms 22 are used to model the systems as mentioned in the previous paragraph.…”