Artificial Neural Networks for Pyrolysis, Thermal Analysis, and Thermokinetic Studies: The Status Quo

Thermogravimetric experiments based prediction of biomass pyrolysis behavior: A comparison of typical machine learning regression models in Scikit-learn

A variety of machine learning (ML) models have been extensively utilized in predicting biomass pyrolysis owing to their prowess in deciphering complex non-linear relationships between inputs and outputs, but there is still a lack of consensus on the optimal methods. This study elaborates on the development, optimization, and evaluation of three ML methodologies, namely, artificial neural networks, random forest (RF), and support vector machines, aimed to determine the optimal model for accurate prediction of biomass pyrolysis behavior using thermogravimetric data. This work assesses the utility of thermal data derived from these models in the computation of kinetic and thermodynamic parameters, alongside an analysis of their statistical performance. Eventually, the RF model exhibits superior physical interpretability and the least discrepancy in predicting kinetic and thermodynamic parameters. Furthermore, a feature importance analysis conducted within the RF model framework quantitatively reveals that temperature and heating rate account for 98.5 % and 1.5 %, respectively.

Improving ADMET Prediction Accuracy for Candidate Drugs: Factors to Consider in QSPR Modeling Approaches

Quantitative Structure-Property Relationship (QSPR) employs mathematical and statistical methods to reveal quantitative correlations between the pharmacokinetics of compounds and their molecular structures, as well as their physical and chemical properties. QSPR models have been widely applied in the prediction of drug absorption, distribution, metabolism, excretion, and toxicity (ADMET). However, the accuracy of QSPR models for predicting drug ADMET properties still needs improvement. Therefore, this paper comprehensively reviews the tools employed in various stages of QSPR predictions for drug ADMET. It summarizes commonly used approaches to building QSPR models, systematically analyzing the advantages and limitations of each modeling method to ensure their judicious application. We provide an overview of recent advancements in the application of QSPR models for predicting drug ADMET properties. Furthermore, this review explores the inherent challenges in QSPR modeling while also proposing a range of considerations aimed at enhancing model prediction accuracy. The objective is to enhance the predictive capabilities of QSPR models in the field of drug development and provide valuable reference and guidance for researchers in this domain.

Study on Thermal Degradation Processes of Polyethylene Terephthalate Microplastics Using the Kinetics and Artificial Neural Networks Models

Because of its slow rate of disintegration, plastic debris has steadily risen over time and contributed to a host of environmental issues. Recycling the world’s increasing debris has taken on critical importance. Pyrolysis is one of the most practical techniques for recycling plastic because of its intrinsic qualities and environmental friendliness. For scale-up and reactor design, an understanding of the degradation process is essential. Using one model-free kinetic approach (Friedman) and two model-fitting kinetic methods (Arrhenius and Coats-Redfern), the thermal degradation of Polyethylene Terephthalate (PET) microplastics at heating rates of 10, 20, and 30 °C/min was examined in this work. Additionally, a powerful artificial neural network (ANN) model was created to forecast the heat deterioration of PET MPs. At various heating rates, the TG and DTG thermograms from the PET MPs degradation revealed the same patterns and trends. This showed that the heating rates do not impact the decomposition processes. The Friedman model showed activation energy values ranging from 3.31 to 8.79 kJ/mol. The average activation energy value was 1278.88 kJ/mol from the Arrhenius model, while, from the Coats-Redfern model, the average was 1.05 × 104 kJ/mol. The thermodynamics of the degradation process of the PET MPs by thermal treatment were all non-spontaneous and endergonic, and energy was absorbed for the degradation. It was discovered that an ANN, with a two-layer hidden architecture, was the most effective network for predicting the output variable (mass loss%) with a regression coefficient value of (0.951–1.0).

Thermal stability of emerging N6-type energetic materials: kinetic modeling of simultaneous thermal analysis data to explain sensitivity trends

A number of new high-performing energetic materials possess explosophoric functionalities, high nitrogen content, and fused heterocyclic blocks. Two representatives of these materials have been synthesized recently, namely, 1,2,9,10-tetranitrodipyrazolo[1,5-d:5',1'-f][1,2,3,4]-tetrazine (1) and 2,9-dinitrobis([1,2,4]triazolo)[1,5-d:5',1'-f][1,2,3,4]tetrazine (2). The thermal stability of these energetic materials bearing the N-N-N = N-N-N fragment and three closely related compounds has been investigated for the first time. The thermal decomposition process of analyzed compounds was complicated by the appearance of the liquid phase, sublimation of the material, and autocatalysis by reaction products. In contrast to the traditional approach to the kinetic modeling based on data from either TGA or DSC, we use both signals' data measured at the same time and perform the joint kinetic analysis using the model-fitting technique to obtain the pertinent kinetic description of the process. Of the analyzed materials, 1 and 2 show the lowest thermal stability in melt with a characteristic rate constant of 2.6 × 10^-3 s^-1 at 250 °C. The kinetic parameters and calculated detonation performance data were used in the model to describe the mechanical sensitivity. The model output and the experimental friction sensitivity data show a respectable agreement, but more data are required to draw firm conclusions. In general, the provided thermal stability and kinetic data can be used for thermal response and storage modeling of these new N6-type energetic materials. The developed thermokinetic approach, joint model-fitting of several thermal analysis signals, can be applied to other complex thermally induced processes to increase the value and credibility of the kinetic findings.

Artificial Neural Network Study on the Pyrolysis of Polypropylene with a Sensitivity Analysis

Among machine learning (ML) studies, artificial neural network (ANN) analysis is the most widely used technique in pyrolysis research. In this work, the pyrolysis of polypropylene (PP) polymers was established using a thermogravimetric analyzer (TGA) with five sets of heating rates (5-40 K min^-1). TGA data was used to exploit an ANN network by achieving a feed-forward backpropagation optimization technique in order to predict the weight-left percentage. Two important ANN model input variables were identified as the heating rate (K min^-1) and temperature (K). For the range of TGA values, a 2-10-10-1 network with two hidden layers (Logsig-Tansig) was concluded to be the best structure for predicting the weight-left percentage. The ANN demonstrated a good agreement between the experimental and calculated values, with a high correlation coefficient (R) of greater than 0.9999. The final network was then simulated with the new input data set for effective performance. In addition, a sensitivity analysis was performed to identify the uncertainties associated with the relationship between the output and input parameters. Temperature was found to be a more sensitive input parameter than the heating rate on the weight-left percentage calculation.

Prediction and Construction of Energetic Materials Based on Machine Learning Methods

Energetic materials (EMs) are the core materials of weapons and equipment. Achieving precise molecular design and efficient green synthesis of EMs has long been one of the primary concerns of researchers around the world. Traditionally, advanced materials were discovered through a trial-and-error processes, which required long research and development (R&D) cycles and high costs. In recent years, the machine learning (ML) method has matured into a tool that compliments and aids experimental studies for predicting and designing advanced EMs. This paper reviews the critical process of ML methods to discover and predict EMs, including data preparation, feature extraction, model construction, and model performance evaluation. The main ideas and basic steps of applying ML methods are analyzed and outlined. The state-of-the-art research about ML applications in property prediction and inverse material design of EMs is further summarized. Finally, the existing challenges and the strategies for coping with challenges in the further applications of the ML methods are proposed.

Catalytic Pyrolysis of PET Polymer Using Nonisothermal Thermogravimetric Analysis Data: Kinetics and Artificial Neural Networks Studies

This paper presents the catalytic pyrolysis of a constant-composition mixture of zeolite β and polyethylene terephthalate (PET) polymer by thermogravimetric analysis (TGA) at different heating rates (2, 5, 10, and 20 K/min). The thermograms showed only one main reaction and shifted to higher temperatures with increasing heating rate. In addition, at constant heating rate, they moved to lower temperatures of pure PET pyrolysis when a catalyst was added. Four isoconversional models, namely, Kissinger−Akahira−Sunose (KAS), Friedman, Flynn−Wall−Qzawa (FWO), and Starink, were applied to obtain the activation energy (Ea). Values of Ea acquired by these models were very close to each other with average value of Ea = 154.0 kJ/mol, which was much lower than that for pure PET pyrolysis. The Coats−Redfern and Criado methods were employed to set the most convenient solid-state reaction mechanism. These methods revealed that the experimental data matched those obtained by different mechanisms depending on the heating rate. Values of Ea obtained by these two models were within the average values of 157 kJ/mol. An artificial neural network (ANN) was utilized to predict the remaining weight fraction using two input variables (temperature and heating rate). The results proved that ANN could predict the experimental value very efficiently (R2 > 0.999) even with new data.

Predicting Pyrolysis of a Wide Variety of Petroleum Coke Using an Independent Parallel Reaction Model and a Backpropagation Neural Network

In this work, the pyrolysis behavior and gaseous products of petroleum coke were investigated by nonisothermal thermogravimetric analysis (TGA) and thermogravimetry-mass spectrometry (TG-MS). Then, the pyrolysis kinetics of six kinds of petroleum coke (Fushun (FS), Fuyu (FY), Wuhan (WH), Zhenhai (ZH), Qilu (QL), and Shijiazhuang (SJZ)) were determined by an independent parallel reaction (IPR) model, and the kinetic parameters (activation energy and preexponential factor) were obtained. In addition, an efficient backpropagation neural network (BPNN) was developed to predict the thermal data of six kinds of petroleum coke. The BPNN-predicted thermal data were used to calculate the kinetic parameters based on the IPR model, and the results were compared with the ones calculated using experimental data. The results showed that the pyrolysis process of six kinds of petroleum coke was divided into three stages, of which stage II (250-900 °C) had the significant mass loss, corresponding to the devolatilization of petroleum coke. MS fragmented ion intensity analysis indicated that the main pyrolysis products were methane CH _x (m/z = 13, 14, 15, and 16), aliphatic hydrocarbon C₃H₅, H₂, CO, CO₂, and H₂O. The thermal data predicted by the IPR, BPNN, and BPNN-IPR (BPNN combined with IPR) models were in good agreement with the experimental data. Most importantly, it was concluded that the BPNN-predicted data can be further applied to calculate the kinetic parameters using the IPR kinetic model.

Thermal Degradation Process of Ethinylestradiol—Kinetic Study

The present study reports the results obtained after the analysis of the thermal stability and decomposition kinetics of widely used synthetic derivative of estradiol, ethinylestradiol (EE), as a pure active pharmaceutical ingredient. As investigational tools, Fourier transformed infrared spectroscopy (FTIR), thermal analysis, and decomposition kinetics modeling of EE were employed. The kinetic study was realized using three kinetic methods, namely Kissinger, Friedman, and Flynn-Wall-Ozawa. The results of the kinetic study are in good agreement, suggesting that the main decomposition process of EE that takes place in the 175–375 °C temperature range is a single-step process, invariable during the modification of heating rate of the sample.

Application of computational approach in plastic pyrolysis kinetic modelling: a review

During the past decade, pyrolysis routes have been identified as one of the most promising solutions for plastic waste management. However, the industrial adoption of such technologies has been limited and several unresolved blind spots hamper the commercial application of pyrolysis. Despite many years and efforts to explain pyrolysis models based on global kinetic approaches, recent advances in computational modelling such as machine learning and quantum mechanics offer new insights. For example, the kinetic and mechanistic information about plastic pyrolysis reactions necessary for scaling up processes is unravelling. This selective literature review reveals some of the foundational knowledge and accurate views on the reaction pathways, product yields, and other features of pyrolysis created by these new tools. Pyrolysis routes mapped by machine learning and quantum mechanics will gain more relevance in the coming years, especially studies that combine computational models with different time and scale resolutions governed by “first principles.” Existing research suggests that, as machine learning is further coupled to quantum mechanics, scientists and engineers will better predict products, yields, and compositions, as well as more complicated features such as ideal reactor design.