Integrating the physics with data analytics for the hybrid modeling of the granulation process

A hybrid model based on physical and data interpretations to investigate the high shear granulation (HSG) process is proposed. This model integrates three separate component models, namely, a computational fluid dynamics model, a population balance model, and a radial basis function model, through an iterative procedure. The proposed hybrid model is shown to provide the required understanding of the HSG process, and to also accurately predict the properties of the granules. Furthermore, a new fusion model based on integrating fuzzy logic theory and the Dempster‐Shafer theory is also developed. The motivation for such a new modeling framework stems from the fact that integrating predictions from models which are elicited using different paradigms can lead to a more robust and accurate topology. As a result, significant improvements in prediction performance have been achieved by applying the proposed framework when compared to single models. © 2017 American Institute of Chemical Engineers AIChE J, 2017     

Model based control of a high purity distillation column

In this paper, the problem of dual product composition control of an industrial high purity distillation column, a deisohexanizer (DIH), is addressed using a Generic Model Control framework. A dynamic simulation of the DIH was performed for preliminary studies of the performance of different controller strategies/algorithms. The performance of Generic Model Control incorporating different process models was studied. Process models are presented ranging from simple first order approximations to mechanistic short cut distillation models where a tradeoff between model complexity and model adaptivity is investigated. The different controllers were implemented and compared using a dynamic simulation of an industrial deisohexanizer (DIH) to select the best condidate controller. A controller using a nonlinear process model emerged as the best controller and was implemented on the actual process, resulting in improved performance over the original controller. Simulation results and industrial plant data are presented.     

The Potential of Hybrid Mechanistic/Data-Driven Approaches for Reduced Dynamic Modeling: Application to Distillation Columns

Extensive literature has considered reduced, but still highly accurate, nonlinear dynamic process models, particularly for distillation columns. Nevertheless, there is a need for continuing research in this field. Herein, opportunities from the integration of machine learning into existing reduction approaches are discussed. First, key concepts for dynamic model reduction and their limitations are briefly reviewed. Afterwards, promising model structures for reduced hybrid mechanistic/data-driven models are outlined. Finally, crucial future challenges as well as promising research perspectives are presented.     

Identification of semi-parametric hybrid process models

Hybrid models are mathematical models that comprise both mechanistic and black-box or data-driven components. Typically, the parameters in the mechanistic part of a hybrid model (if any) are assumed to be known. However in this research, a two-level approach is proposed for the identification of hybrid models where some parameters in the mechanistic part of the model are unknown. At the first level, the black-box component is identified using a regularization method with given values for the regularization and mechanistic parameters. At the second level, the regularization and mechanistic parameters are determined simultaneously and optimized according to a specific criterion placed on the predictive performance of the hybrid model. This approach is tested through the modelling of a toluene nitration process, where a support vector machine (SVM) model is used to represent the chemical kinetics, with the mass transfer-related mechanistic parameters being estimated simultaneously. The case study shows that good results can be obtained in terms of both the prediction of the process variables of interest and the estimates of the mechanistic parameters, when the measurement error in the training data is small whilst when the magnitude of the measurement error increases, the accuracy of the estimates of the mechanistic parameters decreases. However, the predictive performance of the resulting hybrid model in the latter case is still acceptable, and can be much better than that attained from the application of a pure black-box model under certain extrapolation conditions.     

Hybrid method integrating machine learning and particle swarm optimization for smart chemical process operations

Modeling and optimization is crucial to smart chemical process operations.However,a large number of nonlinearities must be considered in a typical chemical process according to complex unit operations,chemical reactions and separations.This leads to a great challenge of implementing mechanistic models into industrial-scale problems due to the resulting computational complexity.Thus,this paper presents an efficient hybrid framework of integrating machine learning and particle swarm optimization to overcome the aforementioned difficulties.An industrial propane dehydrogenation process was carried out to demonstrate the validity and efficiency of our method.Firstly,a data set was generated based on process mechanistic simulation validated by industrial data,which provides sufficient and reasonable samples for model training and testing.Secondly,four well-known machine learning methods,namely,K-nearest neighbors,decision tree,support vector machine,and artificial neural network,were compared and used to obtain the prediction models of the processes operation.All of these methods achieved highly accurate model by adjusting model parameters on the basis of high-coverage data and properly features.Finally,optimal process operations were obtained by using the particle swarm optimization approach.     

USING POWER-LAW RATE EXPRESSION PARAMETERS FOR DISCRIMINATION AMONG MECHANISTIC MODELS IN RATE DATA REGRESSION

Langmuir-Hinshelwood (or Hougen-Watson) type rate expressions are most often used in modeling reaction rate data. In cases when there are tens of possible rival models, effective discrimination between them requires that most of the inadequate models be discarded as early as possible in the discrimination process.

In this paper, the relationship between power-law and Hougen and Watson type rate expressions was studied. It was found that there is a clear mathematical connection between the two types of rate expressions. This connection can be utilized in order to discriminate between feasible and infeasible models, using only the numerical values of the power-law rate expression parameters. This way most of the inadequate mechanisms can be discarded after fitting the data to a single (power-law) model.

The 95% confidence intervals of the parameters have proven to be key statistical variables in determining the adequacy of both power-law and mechanistic models. Using the data and results of Hougen and Watson (1947), it is shown that they rejected valid mechanisms and accepted invalid ones because they did not take into account the confidence intervals on the parameters.     

Integration of machine learning and first principles models

Model building and parameter estimation are traditional concepts widely used in chemical, biological, metallurgical, and manufacturing industries. Early modeling methodologies focused on mathematically capturing the process knowledge and domain expertise of the modeler. The models thus developed are termed first principles models (or white-box models). Over time, computational power became cheaper, and massive amounts of data became available for modeling. This led to the development of cutting edge machine learning models (black-box models) and artificial intelligence (AI) techniques. Hybrid models (gray-box models) are a combination of first principles and machine learning models. The development of hybrid models has captured the attention of researchers as this combines the best of both modeling paradigms. Recent attention to this field stems from the interest in explainable AI (XAI), a critical requirement as AI systems become more pervasive. This work aims at identifying and categorizing various hybrid models available in the literature that integrate machine-learning models with different forms of domain knowledge. Benefits such as enhanced predictive power, extrapolation capabilities, and other advantages of combining the two approaches are summarized. The goal of this article is to consolidate the published corpus in the area of hybrid modeling and develop a comprehensive framework to understand the various techniques presented. This framework can further be used as the foundation to explore rational associations between several models.     

Hybrid semi-parametric modeling in process systems engineering: Past,present and future

Hybrid semi-parametric models consist of model structures that combine parametric and nonparametric submodels based on different knowledge sources. The development of a hybrid semi-parametric model can offer several advantages over traditional mechanistic or data-driven modeling, as reviewed in this paper. These advantages, such as broader knowledge base, transparency of the modeling approach and cost-effective model development, have been widely recognized, not only in academia but also in the industry.In this paper, the most common hybrid semi-parametric modeling and parameter identification techniques are revisited. Applications in the areas of (bio)chemical engineering for process monitoring, control, optimization, scale-up and model-reduction are reviewed. It is outlined that the application of hybrid semi-parametric techniques does not automatically lead into better results but that rational knowledge integration has potential to significantly improve model-based process operation and design.     

MINK AND MUSKRAT INTERACTION:A STRUCTURAL ANALYSIS

Abstract. There is no unique way of representing a linear vector process, but in practice these processes are often characterized by their ARIMA representations. It is argued that for the purpose of interpretation the choice of canonical form is important, and the chosen form should as closely as possible correspond to existing prior information about the process, if any. This point is demonstrated by reanalysing the well-known mink and muskrat data of Hudson's Bay Company from the year 1848 till 1909 using two separate single-equation transfer function noise models to describe the relationship from mink to muskrat and the feedback link from muskrat to mink. The results thus obtained are from the practical point of view more informative than the corresponding ARIMA models.     

Stability and folding behavior analysis of zinc-finger using simple models

Zinc-fingers play crucial roles in regulating gene expression and mediating protein-protein interactions. In this article, two different proteins (Sp1f2 and FSD-1) are investigated using the Gaussian network model and anisotropy elastic network model. By using these simple coarse-grained methods, we analyze the structural stabilization and establish the unfolding pathway of the two different proteins, in good agreement with related experimental and molecular dynamics simulation data. From the analysis, it is also found that the folding process of the zinc-finger motif is predominated by several factors. Both the zinc ion and C-terminal loop affect the folding pathway of the zinc-finger motif. Knowledge about the stability and folding behavior of zinc-fingers may help in understanding the folding mechanisms of the zinc-finger motif and in designing new zinc-fingers. Meanwhile, these simple coarse-grained analyses can be used as a general and quick method for mechanistic studies of metalloproteins.     

聚合反应过程质量指标的推理估计混合模型

针对聚合反应过程的非线性、时变性和不确定性，提出了一种多类型混联混合推理估计模型。该模型以过程机理知识为基础框架，以各种神经网络和回归辩识模型的计算结果作为混合模型中各子模型或机理模型的过程参数。为了体现过程的多模式集成特点，该混合模型充分利用各种类型模型的不同特性，既保证按照动力学规律描述聚合反应过程特性，又充分利用现场运行和分析的数据，辩识模型结构参数，使所建模型不必完全依赖对过程特性的认识。将该混合模型用于聚丙烯腈生产过程质量指标的推理估计，现场应用效果证明了这种模型的优良性能。     

Generalized autoregressive moving average models with GARCH errors

One of the important and widely used classes of models for non-Gaussian time series is the generalized autoregressive model average models (GARMA), which specifies an ARMA structure for the conditional mean process of the underlying time series. However, in many applications one often encounters conditional heteroskedasticity. In this article, we propose a new class of models, referred to as GARMA-GARCH models, that jointly specify both the conditional mean and conditional variance processes of a general non-Gaussian time series. Under the general modeling framework, we propose three specific models, as examples, for proportional time series, non-negative time series, and skewed and heavy-tailed financial time series. Maximum likelihood estimator (MLE) and quasi Gaussian MLE are used to estimate the parameters. Simulation studies and three applications are used to demonstrate the properties of the models and the estimation procedures.     

Decision support for the development,simulation and optimization of dynamic process models

Simulation is besides experimentation the major method for designing,analyzing and optimizing chemical processes.The ability of simulations to reflect real process behavior strongly depends on model quality.Validation and adaption of process models are usually based on available plant data.Using such a model in various simulation and optimization studies can support the process designer in his task.Beneath steady state models there is also a growing demand for dynamic models either to adapt faster to changing conditions or to reflect batch operation.In this contribution challenges of extending an existing decision support framework for steady state models to dynamic models will be discussed and the resulting opportunities will be demonstrated for distillation and reactor examples.     

A hybrid science-guided machine learning approach for modeling chemical processes: A review

This study presents a broad perspective of hybrid process modeling combining the scientific knowledge and data analytics in bioprocessing and chemical engineering with a science-guided machine learning (SGML) approach. We divide the approach into two major categories: ML complements science, and science complements ML. We review the literature relating to the hybrid SGML approach, and propose a systematic classification of hybrid SGML models. For applying ML to improve science-based models, we present expositions of direct serial and parallel hybrid modeling and their combinations, inverse modeling, reduced-order modeling, quantifying uncertainty in the process and even discovering governing equations of the process model. For applying scientific principles to improve ML models, we discuss the science-guided design, learning and refinement. For each subcategory, we identify its requirements, strengths, and limitations, together with their published and potential applications. We also present several examples to illustrate different hybrid SGML methodologies for modeling chemical processes.     

基于混合模型的气流床气化炉建模

为了提高在煤质改变及工艺参数波动条件下气流床气化炉出口结果的预测精度,分别采用机理模型、广义回归神经网络（GRNN）模型以及混合模型对气化炉进行建模,其中混合模型由GRNN模型和机理模型构建,结合两种不同的煤样对三种模型的预测结果进行分析。结果表明：三种模型均可以较好地对气化过程进行模拟;其中在煤种固定的情况下混合模型关于气化温度和CO、CO₂及H₂含量的预测误差为0.18%和0.25%、1.72%及0.43%,与机理模型和GRNN模型相比误差更小;在煤种改变的情况下混合模型关于出口气体结果的预测最接近实际生产数据,误差为0.81%和0.11%、2.53%及0.42%。证明混合模型在煤种改变及工艺参数波动条件下可以有效地对气化过程进行模拟,在很大程度上提高了机理模型和GRNN模型的预测精度。     

不同结构纺织复合材料准静态侵彻实验分析及有限元模拟

本文研究了两种不同结构三维结构纺织复合材料——三维正交机织玻璃纤维,不饱和树脂复合材料和双轴向纬编针织玻璃纤维／不饱和树脂复合材料在MTS材料试验机上的准静态侵彻测试。以纯铝MTS实验数据为标定,分析了准静态侵彻载荷一位移曲线及其破坏机理,比较了不同结构纺织复合材料以及纯铝的位移一载荷曲线,由此计算得到位移与吸能关系曲线。同时根据复合材料各自织物中纤维束排列及织物成型特点,分别建立了复合材料的细观结构模型和单胞模型。编写用户子程序（VUMAT）,用ABAQUS软件进行了有限元模拟。结果表明：三维纺织复合材料各自损伤结果和载荷-位移曲线与实验结果吻合较好,证明有限元的有效性。三维正交结构复合材料抗侵彻能力优于双轴向纬编针织复合材料,但是破坏过程中其抗侵彻能力幅值变化差异大,没有针织复合材料抗侵彻能力稳定。     

Machine learning-based model predictive control of hybrid dynamical systems

This article presents a machine learning-based model predictive control (MPC) scheme for stabilization of hybrid dynamical systems, for which the evolution of states exhibits both continuous and discrete dynamics described by differential and difference equations, respectively. We first present the development of two recurrent neural networks (RNNs) for approximating continuous- and discrete-time dynamics of hybrid dynamical systems, respectively, and then construct a unified hybrid RNN by integrating the two RNN models to capture both continuous and discrete dynamics. The hybrid RNN is used as the prediction model in Lyapunov-based MPC (RNN-LMPC), under which closed-loop stability of hybrid dynamical systems is established. Finally, two case studies including a bouncing ball example and a chemical process are utilized to illustrate the open- and closed-loop performance of the proposed RNN-LMPC scheme.     

Modelling concepts and their relation to the scale of the problem

Models are often discussed in terms of how deterministic or mechanistic they are. These attributes are usually assumed to be independent of the scale at which the model is used, but they are not. Scale also influences the validation and parameterization of models. These issues are discussed with reference to the papers submitted to the conference in the context of the scale diagram of Hoosbeek and Bryant [21].     

Towards controlled cationic polymer growth from inorganic oxide defects: Directing the mechanism of polystyrene grafting from γ-irradiated silica

Most studies on surface-initiated controlled polymerizations for the synthesis of polymeric covalent organic-inorganic hybrid materials focus on chemical methods requiring specific modifications of the inorganic substrate. Few mechanistically-aware approaches have been undertaken towards exploiting the reactivity of defects induced by physical techniques such as ionizing radiations or UV–Vis light. Within this framework, we take grafted polymerization of styrene from γ-irradiated silica as a mechanistic testing ground where para- and diamagnetic silica defects are present, and polymerization proceeds through both radical and cationic mechanisms, resulting in a bimodal molecular weight distribution. We show that these mechanistic intricacies can be sorted out by resorting to the chemical arsenal developed in the last decades for controlled polymerizations. Specifically, we obtained a silica-polystyrene grafted material by cationic grafting from at 30 °C, a unimodal molecular weight distribution, and a relatively high molecular weight (Mn = 7.4 kDa) with a PDI of 1.68.