Artificial Neural Network Modeling of Hydrodynamics of Liquid‐Solid Circulating Fluidized Beds

Solids holdup and solids circulation rate are the two important hydrodynamic variables affected by process conditions. These two variables have a significant influence on the performance of a liquid‐solid circulating fluidized bed (LSCFB). An artificial neural network (ANN) methodology was developed and simulated to predict the performance of the LSCFB for the experimental dataset collected under various process conditions. Different statistical parameters were applied to evaluate the prominent and unique characteristic features of the ANN‐predicted parameters. The ANN model successfully predicted the experimental observations and captured the actual nonlinear behavior noticed during the experiments. Model validation confirmed that this data‐driven technique can be used to model such nonlinear systems.     

Modeling and optimization of cross-flow ultrafiltration using hybrid neural network-genetic algorithm approach

Precise modeling flux decline under various operating parameters in cross-flow ultrafiltration (UF) of oily wastewaters and afterward, employing an appropriate optimization algorithm in order to optimize operating parameters involved in the process model result in attaining desired permeate flux, is of fundamental great interest from an economical and technical point of view. Accordingly, this current research proposed a hybrid process modeling and optimization based on computational intelligence paradigms where the combination of artificial neural network (ANN) and genetic algorithm (GA) meets the challenge of specified-objective based on two steps: first the development of bio-inspired approach based on ANN, trained, validated and tested successfully with experimental data collected during the polyacrylonitrile (PAN) UF process to treat the oily wastewater of Tehran refinery in a laboratory scale in which the model received feed temperature (T), feed pH, trans-membrane pressure (TMP), cross-flow velocity (CFV), and filtration time as inputs; and gave permeate flux as an output. Subsequently, the 5-dimensional input space of the ANN model portraying process input variables was optimized by applying GA, with a view to realizing maximum or minimum process output variable. The results obtained validate the estimates of the ANN–GA technique with a good accuracy. Finally, the relative importance of the controllable operation factors on flux decline is determined by applying the various correlation statistic techniques. According to the result of the sensitivity analysis based on the correlation coefficient, the filtration time was the most significant one, followed by T, CFV, feed pH and TMP.     

Biodegradable iron chelate for H2S abatement: Modeling and optimization using artificial intelligence strategies

A batch reactor process for the abatement of a common pollutant, namely, H₂S using Fe³⁺-malic acid chelate (Fe³⁺-MA) catalyst has been developed. Further, process modeling and optimization was conducted in the three stages with a view to maximize the H₂S conversion: (i) sensitivity analysis of process inputs was performed to select the most influential process operating variables and parameters, (ii) an artificial neural network (ANN)-based data-driven process model was developed using the influential process variables and parameters as model inputs, and H₂S conversion (%) as the model output, and (iii) the input space of the ANN model was optimized using the artificial immune systems (AIS) formalism. The AIS is a recently proposed stochastic nonlinear search and optimization method based on the human biological immune system and has been introduced in this study for chemical process optimization. The AIS-based optimum process conditions have been compared with those obtained using the genetic algorithms (GA) formalism. The AIS-optimized process conditions leading to high (≈97%) H₂S conversion, were tested experimentally and the results obtained thereby show an excellent match with the AIS-maximized H₂S conversion. It was also observed that the AIS required lesser number of generations and function evaluations to reach the convergence when compared with the GA.     

Dynamic crude oil fouling prediction in industrial preheaters using optimized ANN based moving window technique

The objective of this paper is to develop and validate a reliable, efficient and robust artificial neural network (ANN) model for online monitoring and prediction of crude oil fouling behavior for industrial shell and tube heat exchangers. To explore the complex dynamics of fouling, a new modeling strategy based on moving-window neural network approach is proposed. The essential character of this modeling approach is online updating of the ANN model whenever a new data block is available, so that it can effectively capture the slowly changing of process dynamics. The results of these models have been compared with appropriate sets of experimental data. The mean relative errors (MRE) of training and prediction subsets were about 6.61% and 8.06%, respectively. Since the data extraction in the refinery was performed every 2 h, the modeling approach led to an MRE of about 8% for fouling rate prediction of the next 50 h.     

Disaggregation-aggregation based model reduction for refinery-wide optimization

In this paper, reduced nonlinear refinery models are developed by generating and using input-output data from a process simulator. In particular, rigorous process models of continuous catalytic reformer (CCR) and naphtha splitter units are used for generating the data. To deal with complexity associated with large amounts of data, that is usually available in the refineries, a disaggregation-aggregation based approach is presented. The data is split (disaggregation) into smaller subsets and reduced artificial neural network (ANN) models are obtained for each of the subset. These ANN models are then combined (aggregation) to obtain an ANN model which represents all the data originally generated. The disaggregation step can be carried out within a parallel computing platform. Refinery optimization studies are carried out to demonstrate the applicability and the usefulness of the proposed model reduction approach.     

A new weighted optimal combination of ANNs for catalyst design and reactor operation: Methane steam reforming studies

Catalyst design and evaluation is a multifactorial multiobjective optimization problem and the absence of well‐defined mechanistic relationships between wide ranging input‐output variables has stimulated interest in the application of artificial neural network for the analysis of the large body of empirical data available. However, single ANN models generally have limited predictive capability and insufficient to capture the broad range of features inherent in the voluminous but dispersed data sources. In this study, we have employed a Fibonacci approach to select optimal number of neurons for the ANN architecture followed by a new weighted optimal combination of statistically‐derived candidate ANN models in a multierror sense. Data from 200 cases for catalytic methane steam reforming have been used to demonstrate the veracity and robustness of the integrated ANN modeling technique. © 2011 American Institute of Chemical Engineers AIChE J, 58: 2412–2427, 2012     

Application of equation of state and artificial neural network to prediction of volumetric properties of polymer melts

The modified ISM EOS and artificial neural network (ANN) methods were used to predict the PVT behavior of polymer melts containing polystyrene (PS), polycarbonate bisphenol-A (PC), polyvinylidene fluoride (PVDF), polypropyleneglycol (PPG), polyethylene glycol (PEG), polypropylene (PP), poly vinylchloride's properties (PVC), poly(1-butene) (PB1), polycaprolactone (PCL), polyethylene (PE) and polyvinyl methyl ether (PVME) over the entire range of available temperature and pressure. The obtained results show that the modified ISM EOS and ANN models have good agreement with the experimental data with mean average absolute deviation of 0.58% and 0.20%, respectively.     

Comparative Modeling and Analysis of the Mass Transfer Coefficient in a Turbulent Bed Contactor using Artificial Neural Network and Adaptive Neuro-Fuzzy Inference Systems

This study deals with predicting the gas film volumetric mass transfer coefficient (K_oga) in a turbulent bed contactor (TBC), using both artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS) techniques. The networks have been trained and evaluated with the experimental data available in the literature. Input variables to the networks are process variables such as gas and liquid phase concentration, gas and liquid superficial velocities and also specific area of packings. The results obtained the ability of developed ANN and ANFIS for prediction of K_oga. Although it was observed that both ANN and ANFIS models provided a good statistical prediction in terms of coefficient of determination (R²), mean relative error (MRE) and root mean square error (RMSE), the accuracies of ANFIS predictions were better than those of ANN predictions.     

Principal component analysis coupled with artificial neural networks-a combined technique classifying small molecular structures using a concatenated spectral database

In this paper we present several expert systems that predict the class identity of the modeled compounds, based on a preprocessed spectral database. The expert systems were built using Artificial Neural Networks (ANN) and are designed to predict if an unknown compound has the toxicological activity of amphetamines (stimulant and hallucinogen), or whether it is a nonamphetamine. In attempts to circumvent the laws controlling drugs of abuse, new chemical structures are very frequently introduced on the black market. They are obtained by slightly modifying the controlled molecular structures by adding or changing substituents at various positions on the banned molecules. As a result, no substance similar to those forming a prohibited class may be used nowadays, even if it has not been specifically listed. Therefore, reliable, fast and accessible systems capable of modeling and then identifying similarities at molecular level, are highly needed for epidemiological, clinical, and forensic purposes. In order to obtain the expert systems, we have preprocessed a concatenated spectral database, representing the GC-FTIR (gas chromatography-Fourier transform infrared spectrometry) and GC-MS (gas chromatography-mass spectrometry) spectra of 103 forensic compounds. The database was used as input for a Principal Component Analysis (PCA). The scores of the forensic compounds on the main principal components (PCs) were then used as inputs for the ANN systems. We have built eight PC-ANN systems (principal component analysis coupled with artificial neural network) with a different number of input variables: 15 PCs, 16 PCs, 17 PCs, 18 PCs, 19 PCs, 20 PCs, 21 PCs and 22 PCs. The best expert system was found to be the ANN network built with 18 PCs, which accounts for an explained variance of 77%. This expert system has the best sensitivity (a rate of classification C = 100% and a rate of true positives TP = 100%), as well as a good selectivity (a rate of true negatives TN = 92.77%). A comparative analysis of the validation results of all expert systems is presented, and the input variables with the highest discrimination power are discussed.     

Application of artificial neural networks for the optimal design of sheet molding compound (SMC) compression molding

A sequential design optimization scheme based on artificial neural networks (ANN) is proposed. It is a combination of an ANN model and a nonlinear programming algorithm. The proposed scheme is implemented with network training, optimization, and sheet molding compound (SMC) process simulation in a closed loop. A “cyclic coordinate search” technique is employed to initiate the optimization process, to collect training data for the neural network model, and to perform a preliminary design sensitivity analysis. Emphasis is placed on the development of an integrated, automatic optimization-simulation design tool that does not rely on the designer's experience and interpretation. Testing results based on the design of heating channels in an SMC compression molding tool show that the optimal design can be achieved with fewer data points than other methods, such as factorial design. The efficiency of the ANN method would be greater as the number of design variables grows.     

Artificial neural network simulation of rare earths solvent extraction equilibrium data

Separation of high purity rare earth elements from their mixed oxides, obtained from monazite or xenotime, requires multiple stages of separation by circuits incorporating one or more solvents. The separation factors being small, a large number of counter-current stages become necessary. Process development, analysis, optimization and control of rare earths are a complex task. Computer simulation provides useful tools in this area. Application of artificial neural networks (ANN) for simulation of equilibrium data in solvent extraction of rare earths is described in this paper. The back propagation ANN model has been used. The input neurons correspond to the system state variables such as equilibrium concentration and acidity. The partitioning of the metal ion into the two immiscible phases involved in solvent extraction is measured in terms of distribution ratio D. The model predicts the D value under varying process conditions. Comparison of ANN with conventional models shows that ANN is superior. The average absolute error for ANN model is one-fourth that of the conventional models. The approach has been used, in conjunction with a process simulation model, successfully for industrial process development involving production of high purity neodymium.     

人工神经网络在半水盐酸帕罗西汀溶解度预测中的应用

溶解度的测定与预测对于多晶型体的晶体生长和结晶过程中的多晶型控制至关重要.利用激光监视装置,首次测得了半水盐酸帕罗西汀在不同溶剂体系中的溶解度,共计308组数据;采用多参数人工神经网络模型,随机选取308组数据中的184组数据进行人工神经网络的训练,考察了不同隐含层节点数对神经网络训练效果的影响,得到了优化后的人工神经网络模型,利用剩余的124组数据对优化后的模型进行检测,平均预测误差小于0.7％.预测结果表明,优化后的人工神经网络模型可以胜任半水盐酸帕罗西汀溶解度预测的任务.     

Process modeling and optimization of industrial ethylene oxide reactor by integrating support vector regression and genetic algorithm

This article presents an artificial intelligence‐based process modeling and optimization strategies, namely support vector regression–genetic algorithm (SVR‐GA) for modeling and optimization of catalytic industrial ethylene oxide (EO) reactor. In the SVR‐GA approach, an SVR model is constructed for correlating process data comprising values of operating and performance variables. Next, model inputs describing process operating variables are optimized using Genetic Algorithm (GAs) with a view to maximize the process performance. The GA possesses certain unique advantages over the commonly used gradient‐based deterministic optimization algorithms The SVR‐GA is a new strategy for chemical process modeling and optimization. The major advantage of the strategies is that modeling and optimization can be conducted exclusively from the historic process data wherein the detailed knowledge of process phenomenology (reaction mechanism, kinetics, etc.) is not required. Using SVR‐GA strategy, a number of sets of optimized operating conditions leading to maximized EO production and catalyst selectivity were obtained. The optimized solutions when verified in actual plant resulted in a significant improvement in the EO production rate and catalyst selectivity.     

Identification of the reaction network for the synthesis of adipic acid using machine learning coupling with target factor analysis

This study investigated the reaction network of the oxidation of cyclohexanone with nitric acid through machine learning (ML) model coupling with target factor analysis (TFA). Experiments for the synthesis of adipic acid (AA) were carefully designed and carried out in a microreactor system. An artificial neural network (ANN) model was applied and optimized by training on experimental data. To assess the established ANN model, a comparison between the predicted concentrations of the products and those calculated with the power law kinetic model was implemented. TFA was then performed on both the experimental data and the data simulated by ANN for identifying the candidate reactions and finding out the temperature boundaries. Based on the identification results of the reaction network, the kinetic characteristics of this oxidation process under various operational conditions were further researched through the incremental approach and simultaneous approach.     

基于人工神经网络的橡胶螺杆挤出机智能化设计

通过人工神经网络建立了挤出机的智能化设计模型.该模型可以进行多影响因素下的多目标分析,并应用于橡胶挤出机的结构选型、结构设计、找出最佳的工艺参数等.改进了过去单凭产量来选择螺杆直径的简略的方法.由实验数据和专家经验作为样本训练得到的智能模型,能达到更符合实际情况、综合考虑产量和胶料特性等诸多因素来选择出合理的挤出机直径和挤出机工作转速等多目标的智能化的设计方法.     

Artificial Neural Network Modeling of Osmotic Dehydration Mass Transfer Kinetics of Fruits

Artificial neural network (ANN) models were developed for the prediction of transient moisture loss (ML) and solid gain (SG) in osmotic dehydration of fruits using process kinetics data from the literature. ANN models for ML and SG were developed based on data over a broad range of operating conditions and ten common processing variables: temperature and concentration of osmotic solution, immersion time, initial water and solid content of the fruit, porosity, surface area, characteristic length, solution-to-fruit mass ratio, and agitation level. The trained models were able to accurately predict the outputs with associated regression coefficients (r) of 0.96 and 0.93, respectively, for ML and SG. These ANN models performed much better than those obtained from linear multivariate regression analysis. The large number of process variables and their wide ranges considered along with their easy implementation in a spreadsheet make them very useful and practical for process design and control.     

Successive approximate model based multi-objective optimization for an industrial straight grate iron ore induration process using evolutionary algorithm

Multi-objective optimization of any complex industrial process using first principle computationally expensive models often demands a substantially higher computation time for evolutionary algorithms making it less amenable for real time implementation. A combination of the above-mentioned first principle model and approximate models based on artificial neural network (ANN) successively learnt in due course of optimization using the data obtained from first principle models can be intelligently used for function evaluation and thereby reduce the aforementioned computational burden to a large extent. In this work, a multi-objective optimization task (simultaneous maximization of throughput and Tumble index) of an industrial iron ore induration process has been studied to improve the operation of the process using the above-mentioned metamodeling approach. Different pressure and temperature values at different points of the furnace bed, grate speed and bed height have been used as decision variables whereas the bounds on cold compression strength, abrasion index, maximum pellet temperature and burn-through point temperature have been treated as constraints. A popular evolutionary multi-objective algorithm, NSGA II, amalgamated with the first principle model of the induration process and its successively improving approximation model based on ANN, has been adopted to carry out the task. The optimization results show that as compared to the PO solutions obtained using only the first principle model, (i) similar or better quality PO solutions can be achieved by this metamodeling procedure with a close to 50% savings in function evaluation and thereby computation time and (ii) by keeping the total number of function evaluations same, better quality PO solutions can be obtained.     

Data-driven integrated design of solvents and extractive distillation processes

As property and process models with many variables need to be considered, integrated computer-aided molecular and process design (CAMPD) problems are computationally expensive. An efficient CAMPD approach is proposed for the simultaneous design of solvents and extractive distillation (ED) processes based on a data-driven modeling strategy. First, artificial neural network (ANN)-based process models are trained to replace the physical models conventionally used in CAMPD. Subsequently, optimization is performed to maximize process performance, through which optimal solvent properties and corresponding optimal process parameters are obtained. Then, real solvents approximating the optimal property values are identified from a large solvent database. Rigorous simulations of the ED process are performed to evaluate the performance of the optimal solvents and corresponding process parameters. Further economic evaluation (6.11% lower annual cost compared to the benchmark process) and chemical hazard assessment confirm that acetylacetone is a promising solvent for the ED separation of 1-butene from 1,3-butadiene.     

Prediction of Extrudate Properties Using Artificial Neural Networks

A backpropagation artificial neural network (ANN) model was developed to predict the properties of extrudates generated by extrusion cooking of fish muscle-rice flour blend in a single screw extruder. Experimental data obtained in a previous study on extrudate properties of expansion ratio, bulk density and hardness at different combinations of operating variables of barrel temperature, feed content and feed moisture had been analysed using response surface methodology (RSM). A backpropagation neural network model was implemented in MATLAB and was trained for operating variables (inputs) and for each individual measured extrudate properties expansion ratio ER, bulk density BD and harndess H (outputs). The optimized network indicated that one hidden layer with a learning rate of 0.1, steep descent learning rule, 100 000 epochs and a logistic sigmoid transfer function predicted the extrudate properties better than RSM. The agreement of the ANN model with the experimental values, expressed as sum of squared error values, was 9.8 × 10⁻⁷ for ER, 5.8 × 10⁻² for BD and 3.8 × 10⁻³ for H. The ANN prediction for the optimized process conditions was superior to the RSM values, with percentage errors of +6.06% (ER), +4.08% (BD) and −14.28% (H).     

Multivariate analysis of high through-put adhesively bonded single lap joints

The goal of this research was to experimentally demonstrate the correlations between processing variables (adhesive type, bondline thickness, adherend thickness, surface pretreatment, overflow fillet) and effective strength in adhesively bonded single lap joints. While generalizations between effective strength and individual joint design parameters have been assumed for decades, the multifaceted interplay between parameters is complex and remains difficult to understand. Traditionally reported studies of the adhesive bond strength of single lap joints are often limited in the sample size populations needed to statistically probe concurrent design variables. To overcome sample size limitations a test matrix of 1200 single lap joints, partitioned by 96 unique fabrication conditions, was processed and tested using a workflow protocol orchestrated through a relational database. The enhanced pedigree and integrity enabled by using a relational database centered workflow allowed for multivariate principal component analysis of the joint design parameters, with all experimental data input available for peer audit. The results of this study revealed that the adhesive type biases the remaining joint configuration variables towards more influence with respect to either mechanical load or displacement to failure.