Causal temporal graph attention network for fault diagnosis of chemical processes

Fault detection and diagnosis(FDD)plays a significant role in ensuring the safety and stability of chemical processes.With the development of artificial intelligence(AI)and big data technologies,data-driven approaches with excellent performance are widely used for FDD in chemical processes.However,improved predictive accuracy has often been achieved through increased model complexity,which turns models into black-box methods and causes uncertainty regarding their decisions.In this study,a causal temporal graph attention network(CTGAN)is proposed for fault diagnosis of chemical processes.A chemical causal graph is built by causal inference to represent the propagation path of faults.The attention mechanism and chemical causal graph were combined to help us notice the key variables relating to fault fluctuations.Experiments in the Tennessee Eastman(TE)process and the green ammonia(GA)process showed that CTGAN achieved high performance and good explainability.

Multivariable Decoupling Predictive Control with Input Constraints and Its Application on Chemical Process

A constrained decoupling (generalized predictive control) GPC algorithm is proposed for MIMO (malti-input multi-output) system. This algorithm takes account of all constraints of inputs and their increments. By solving matrix equations, the multi-step predictive decoupling controllers are realized. This algorithm need not solve Diophantine functions, and weakens the cross-coupling of the variables. At last the simulation results demon- strate the effectiveness of this proposed strategy.

Mechanism of Formation of the Ozone Valley over the Tibetan Plateau in Summer-Transport and Chemical Process of Ozone

With the 3D chemical transport model OSLO CTM2, the valley of total column ozone over the Tibetan Plateau in summer is reproduced. The results show that when the ozone valley occurs and develops, the transport process plays the main part in the ozone reduction, but the chemical process partly compensates for the transport process. In the dynamic transport process of ozone, the horizontal transport process plays the main part in the ozone reduction in May, but brings about the ozone increase in June and July. The vertical advective process gradually takes the main role in the ozone reduction in June and July. The effect of convective activities rises gradually so that this effect cannot be overlooked in July, as its magnitude is comparable to that of the net changes. The effect of the gaseous chemical process brings about ozone increases which are more than the net changes sometimes, so the chemical effect is also important.

Dimensionality Reduction with Input Training Neural Network and Its Application in Chemical Process Modelling

Many applications of principal component analysis (PCA) can be found in dimensionality reduction. But linear PCA method is not well suitable for nonlinear chemical processes. A new PCA method based on im-proved input training neural network (IT-NN) is proposed for the nonlinear system modelling in this paper. Mo-mentum factor and adaptive learning rate are introduced into learning algorithm to improve the training speed of IT-NN. Contrasting to the auto-associative neural network (ANN), IT-NN has less hidden layers and higher training speed. The effectiveness is illustrated through a comparison of IT-NN with linear PCA and ANN with experiments. Moreover, the IT-NN is combined with RBF neural network (RBF-NN) to model the yields of ethylene and propyl-ene in the naphtha pyrolysis system. From the illustrative example and practical application, IT-NN combined with RBF-NN is an effective method of nonlinear chemical process modelling.

Computational Mass Transfer Method for Chemical Process Simulation

The recent works on the development of computational mass transfer （CMT） method and its applications in chemical process simulation are reviewed. Some development strategies and challenges in future research are also discussed.

SDG-based Model Validation in Chemical Process Simulation

Signed direct graph (SDG) theory provides algorithms and methods that can be applied directly to chemical process modeling and analysis to validate simulation models, and is a basis for the development of a software environment that can automate the validation activity. This paper is concentrated on the pretreatment of the model validation. We use the validation scenarios and standard sequences generated by well-established SDG model to validate the trends fitted from the simulation model. The results are helpful to find potential problems, assess possible bugs in the simulation model and solve the problem effectively. A case study on a simulation model of boiler is presented to demonstrate the effectiveness of this method.

A Hybrid Improved Genetic Algorithm and Its Application in Dynamic Optimization Problems of Chemical Processes

The solutions of dynamic optimization problems are usually very difficult due to their highly nonlinear and multidimensional nature. 13enetic algorithm （GA） has been proved to be a teasibte method when the gradient is difficult to calculate. Its advantage is that the control profiles at all time stages are optimized simultaneously, but its convergence is very slow in the later period of evolution and it is easily trapped in the local optimum. In this study, a hybrid improved genetic algorithm （HIGA） for solving dynamic optimization problems is proposed to overcome these defects. Simplex method （SM） is used to perform the local search in the neighborhood of the optimal solution. By using SM, the ideal searching direction of global optimal solution could be found as soon as possible and the convergence speed of the algorithm is improved. The hybrid algorithm presents some improvements, such as protecting the best individual, accepting immigrations, as well as employing adaptive crossover and Ganssian mutation operators. The efficiency of the proposed algorithm is demonstrated by solving several dynamic optimization problems. At last, HIGA is applied to the optimal production of secreted protein in a fed batch reactor and the optimal feed-rate found by HIGA is effective and relatively stable.

PCA weight and Johnson transformation based alarm threshold optimization in chemical processes

To alleviate the heavy load of massive alarm on operators, alarm threshold in chemical processes was optimized with principal component analysis(PCA) weight and Johnson transformation in this paper. First, few variables that have high PCA weight factors are chosen as key variables. Given a total alarm frequency to these variables initially, the allowed alarm number for each variable is determined according to their sampling time and weight factors. Their alarm threshold and then control limit percentage are determined successively. The control limit percentage of non-key variables is determined with 3σ method alternatively. Second, raw data are transformed into normal distribution data with Johnson function for all variables before updating their alarm thresholds via inverse transformation of obtained control limit percentage. Alarm thresholds are optimized by iterating this process until the calculated alarm frequency reaches standard level(normally one alarm per minute). Finally,variables and their alarm thresholds are visualized in parallel coordinate to depict their variation trends concisely and clearly. Case studies on a simulated industrial atmospheric-vacuum crude distillation demonstrate that the proposed alarm threshold optimization strategy can effectively reduce false alarm rate in chemical processes.

Identification of abnormal conditions in high-dimensional chemical process based on feature selection and deep learning

Identification of abnormal conditions is essential in the chemical process.With the rapid development of artificial intelligence technology,deep learning has attracted a lot of attention as a promising fault identification method in chemical process recently.In the high-dimensional data identification using deep neural networks,problems such as insufficient data and missing data,measurement noise,redundant variables,and high coupling of data are often encountered.To tackle these problems,a feature based deep belief networks(DBN)method is proposed in this paper.First,a generative adversarial network(GAN)is used to reconstruct the random and non-random missing data of chemical process.Second,the feature variables are selected by Spearman’s rank correlation coefficient(SRCC)from high-dimensional data to eliminate the noise and redundant variables and,as a consequence,compress data dimension of chemical process.Finally,the feature filtered data is deeply abstracted,learned and tuned by DBN for multi-case fault identification.The application in the Tennessee Eastman(TE)process demonstrates the fast convergence and high accuracy of this proposal in identifying abnormal conditions for chemical process,compared with the traditional fault identification algorithms.

A Strategy for Multi-objective Optimization under Uncertainty in Chemical Process Design

In many circumstances, chemical process design can be formulated as a multi-objective optimization （MOO） problem. Examples include bi-objective optimization problems, where the economic objective is maximized and environmental impact is minimized simultaneously. Moreover, the random behavior in the process,property, market fluctuation, errors in model prediction and so on would affect the performance of a process. Therefore, it is essential to develop a MOO methodology under uncertainty. In this article, the authors propose a generic and systematic optimization methodology for chemical process design under uncertainty. It aims at identifying the optimal design from a number of candidates. The utility of this methodology is demonstrated by a case study based on the design of a condensate treatment unit in an ammonia plant.

Adaptive multiscale convolutional neural network model for chemical process fault diagnosis

Intelligent fault recognition techniques are essential to ensure the long-term reliability of manufacturing.Due to the variations in material,equipment and environment,the process variables monitored by sensors contain diverse data characteristics at different time scales or in multiple operating modes.Despite much progress in statistical learning and deep learning for fault recognition,most models are constrained by abundant diagnostic expertise,inefficient multiscale feature extraction and unruly multimode condition.To overcome the above issues,a novel fault diagnosis model called adaptive multiscale convolutional neural network(AMCNN)is developed in this paper.A new multiscale convolutional learning structure is designed to automatically mine multiple-scale features from time-series data,embedding the adaptive attention module to adjust the selection of relevant fault pattern information.The triplet loss optimization is adopted to increase the discrimination capability of the model under the multimode condition.The benchmarks CSTR simulation and Tennessee Eastman process are utilized to verify and illustrate the feasibility and efficiency of the proposed method.Compared with other common models,AMCNN shows its outstanding fault diagnosis performance and great generalization ability.

Novel Control Vector Parameterization Method with Differential Evolution Algorithm and Its Application in Dynamic Optimization of Chemical Processes

Two general approaches are adopted in solving dynamic optimization problems in chemical processes, namely, the analytical and numerical methods. The numerical method, which is based on heuristic algorithms, has been widely used. An approach that combines differential evolution （DE） algorithm and control vector parameteri- zation （CVP） is proposed in this paper. In the proposed CVP, control variables are approximated with polynomials based on state variables and time in the entire time interval. Region reduction strategy is used in DE to reduce the width of the search region, which improves the computing efficiency. The results of the case studies demonstrate the feasibility and efficiency of the oroposed methods.

Recent progress on equation-oriented optimization of complex chemical processes

Process optimization in equation-oriented(EO)modeling environments favors the gradient-based optimization algorithms by their abilities to provide accurate Jacobian matrices via automatic or symbolic differentiation.However,computational inefficiencies including that in initial-point-finding for Newton type methods have significantly limited its application.Recently,progress has been made in using a pseudo-transient(PT)modeling method to address these difficulties,providing a fresh way forward in EO-based optimization.Nevertheless,research in this area remains open,and challenges need to be addressed.Therefore,understanding the state-of-the-art research on the PT method,its principle,and the strategies in composing effective methodologies using the PT modeling method is necessary for further developing EO-based methods for process optimization.For this purpose,the basic concepts for the PT modeling and the optimization framework based on the PT model are reviewed in this paper.Several typical applications,e.g.,complex distillation processes,cryogenic processes,and optimizations under uncertainty,are presented as well.Finally,we identify several main challenges and give prospects for the development of the PT based optimization methods.

Microfluidic field strategy for enhancement and scale up of liquid-liquid homogeneous chemical processes by optimization of 3D spiral baffle structure

Due to the scale effect, the uniform distribution of reagents in continuous flow reactor becomes bad when the channel is enlarged to tens of millimeters. Microfluidic field strategy was proposed to produce high mixing efficiency in large-scale channel. A 3D spiral baffle structure(3SBS) was designed and optimized to form microfluidic field disturbed by continuous secondary flow in millimeter scale Y-shaped tube mixer(YSTM). Enhancement effect of the 3SBS in liquid-liquid homogeneous chemical processes was verified and evaluated through the combination of simulation and experiment. Compared with 1 mm YSTM, 10 mm YSTM with 3SBS increased the treatment capacity by 100 times, shortened the basic complete mixing time by 0.85 times, which proves the potential of microfluidic field strategy in enhancement and scale-up of liquid-liquid homogeneous chemical process.

A New Optimal Control System Design for Chemical Processes

Based on frequency response and convex optimization,a novel optimal control system was developed for chemical processes.The feedforward control is designed to improve the tracking performance of closed loop chemical systems.The parametric model is not required because the system directly utilizes the frequency response of the loop transfer function,which can be measured accurately.In particular,the extremal values of magnitude and phase can be solved according to constrained quadratic programming optimizer and convex optimization.Simulation examples show the effectiveness of the method.The design method is simple and easily adopted in chemical industry.

Dynamic chemical processes on ZnO surfaces tuned by physisorption under ambient conditions

The catalytic properties of non-reducible metal oxides have intrigued continuous interest in the past decades.Often time,catalytic studies of bulk non-reducible oxides focused on their high-temperature applications owing to their weak interaction with small molecules.Hereby,combining ambient-pressure scanning tunneling microscopy(AP-STM),AP X-ray photoelectron spectroscopy(AP-XPS)and density functional theory(DFT)calculations,we studied the activation of CO and CO_(2)on ZnO,a typical nonreducible oxide and major catalytic material in the conversion of C1 molecules.By visualizing the chemical processes on ZnO surfaces at the atomic scale under AP conditions,we showed that new adsorbate structures induced by the enhanced physisorption and the concerted interaction of physisorbed molecules could facilitate the activation of CO and CO_(2)on ZnO.The reactivity of ZnO towards CO could be observed under AP conditions,where an ordered(2×1)–CO structure was observed on ZnO(1010).Meanwhile,chemisorption of CO_(2)on ZnO(1010)under AP conditions was also enhanced by physisorbed CO_(2),which minimizes the repulsion between surface dipoles and causes a(3×1)–CO_(2)structure.Our study has brought molecular insight into the fundamental chemistry and catalytic properties of ZnO surfaces under realistic reaction conditions.

Topology Chemical Process Research on the Condensing and Growing Phase Multi Micro Oil Droplets Based on Hydrodynamics

The hydrodynamic research about the droplet condensing of the multi phase liquid state on the surface of the coal glass and water discusses the deepening process of convex shape curve and the formation of S shape, and puts emphasis on describing the diagram formation method of the later. In the induction period the active diagram of the micro droplet is decided by pH value forming as convex shape diagram or S shape diagram. When pH value is above 4.0, the damage of convex shape diagram cannot be recovered, in that case produce S shape activity diagram. When pH value is equal to or above 12.0, the hard surface with alkali liquid state loses adhesion, so that the micro droplet condensing of the multi phase liquid state stops completely. The research result shows that the water cleaning conditions of getting rid of the oil micro droplets can be decided by the pH value.

Application in soft sensing modeling of chemical process based on K-OPLS method

Aiming at the problem of soft sensing modeling for chemical process with strong nonlinearity and complexity,a soft sensing modeling method based on kernel-based orthogonal projections to latent structures(K-OPLS)is proposed.Orthogonal projections to latent structures(O-PLS)is a general linear multi-variable data modeling method.It can eliminate systematic variations from descriptive variables(input)that are orthogonal to response variables(output).In the framework of O-PLS model,K-OPLS method maps descriptive variables to high-dimensional feature space by using“kernel technique”to calculate predictive components and response-orthogonal components in the model.Therefore,the K-OPLS method gives the non-linear relationship between the descriptor and the response variables,which improves the performance of the model and enhances the interpretability of the model to a certain extent.To verify the validity of K-OPLS method,it was applied to soft sensing modeling of component content of debutane tower base butane(C4),the quality index of the key product output for industrial fluidized catalytic cracking unit(FCCU)and H 2S and SO 2 concentration in sulfur recovery unit(SRU).Compared with support vector machines(SVM),least-squares support-vector machine(LS-SVM),support vector machine with principal component analysis(PCA-SVM),extreme learning machine(ELM),kernel based extreme learning machine(KELM)and kernel based extreme learning machine with principal component analysis(PCA-KELM)methods under the same conditions,the experimental results show that the K-OPLS method has superior modeling accuracy and good model generalization ability.

Dynamic Multi-objective Optimization of Chemical Processes Using Modified BareBones MOPSO Algorithm

Dynamic multi-objective optimization is a complex and difficult research topic of process systems engineering. In this paper,a modified multi-objective bare-bones particle swarm optimization( MOBBPSO) algorithm is proposed that takes advantage of a few parameters of bare-bones algorithm. To avoid premature convergence,Gaussian mutation is introduced; and an adaptive sampling distribution strategy is also used to improve the exploratory capability. Moreover, a circular crowded sorting approach is adopted to improve the uniformity of the population distribution.Finally, by combining the algorithm with control vector parameterization,an approach is proposed to solve the dynamic optimization problems of chemical processes. It is proved that the new algorithm performs better compared with other classic multiobjective optimization algorithms through the results of solving three dynamic optimization problems.

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.