Applications of Artificial Neural Networks in Process Control Applications: A Review

Krämer, Alexander; Morgado‐Dias, Fernando

doi:10.1109/icbea.2018.8471743

Cited by 12 publications

(8 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…There are many methods for developing statistical or data-driven models, such as least square, support vector regression, artificial neural network (ANN), and so on. Among the data-driven models, ANN is the most widely applied framework for developing nonlinear models . Pirdashti et al provided a review on the applications of ANN.…”

Section: Introductionmentioning

confidence: 99%

Energy and Production Efficiency Optimization of an Ethylene Plant Considering Process Operation and Structure

Wang

Zhu

2020

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

Nowadays, optimizing the efficiency of energy and production has become a hot research area. In this paper, a hybrid multiobjective optimization model integrating NSGA-II and the genetic algorithm with artificial neural network is proposed to improve the production efficiency while reducing the total energy consumption of an ethylene plant incorporating process operating variables and structure in the cryogenic separation system. A novel multiobjective mix-integer nonlinear programming (MOMINLP) model is built to obtain key decision variables and the Pareto frontier of multiobjective optimization. Specifically, an accurate model of the heat capacity affected by the temperature and phase changes is established to enhance the estimation of energy requirements. In addition, redundant variables with little effect or even bad influences on multiobjective optimization are removed, which reduces complexity and further improves performance. To verify the performance of the proposed methodology, two case studies concerning optimizing the process operating conditions and structure of an ethylene plant using the proposed MOMINLP are carried out. Simulation results show that the overall profits of the ethylene plant are improved and the energy consumption is effectively reduced by taking the economic and energy cycles into consideration, which indicates that the proposed methodology can provide an effective way to improving production efficiency while reducing the energy consumption.

show abstract

Section: Introductionmentioning

confidence: 99%

Energy and Production Efficiency Optimization of an Ethylene Plant Considering Process Operation and Structure

Wang

Zhu

2020

Ind. Eng. Chem. Res.

View full text Add to dashboard Cite

show abstract

“…Various successful applications of artificial neural networks exist these days. For instance, neural networks are applied in process identification, function approximation, pattern recognition, time series prediction and many other examples through the various fields, as summarized in [1] or in [2], Generally, an artificial neural network is a complex structure with many parameters needed to be selected in order to find the best behavior and results. For example, topology, method of weight initialization, particular activation function selection, and training algorithm implementation.…”

Section: Introductionmentioning

confidence: 99%

Comparison of ReLU and linear saturated activation functions in neural network for universal approximation

Štursa

Doležel

2019

2019 22nd International Conference on Process Control (PC19)

View full text Add to dashboard Cite

Activation functions used in hidden layers directly affect the possibilities for describing nonlinear systems using a feedforward neural network. Furthermore, linear based activation functions are less computationally demanding than their nonlinear alternatives. In addition, feedforward neural networks with linear based activation functions can be advantageously used for control of nonlinear systems, as shown in previous authors' publications. This paper aims to compare two types of linear based functions-symmetric linear saturated function and the rectifier linear unit (ReLU) function as activation functions of the feedforward neural network used for a nonlinear system approximation. Topologies with one hidden layer and the combination of defined quantities of hidden layer neurons in the feedforward neural network are used. Strict criteria are applied for the conditions of the experiments; specifically, the Levenberg-Marquardt algorithm is applied as a training algorithm and the Nguyen-Widrow algorithm is used for the weights and biases initialization. Three benchmark systems are then selected as nonlinear plants for approximation, which should serve as a repeatable source of data for testing. The training data are acquired by the computation of the output as a reaction to a specified colored input signal. The comparison is based on the convergence speed of the training for a fixed value of the error function, and also on the performance over a constant number of epochs. At the end of the experiments, only small differences between the performance of both applied activation functions are observed. Although the symmetric linear saturated activation function provides the lesser median of the final error function value across the all tested numbers of neurons in topologies, the ReLU function seems to be also capable of use as the activation function for nonlinear system modeling.

show abstract

“…Dentre as estruturas que podem ser aplicadas, os controladores MPC (do inglês Model Predictive Control ) e ANN-Fuzzy (baseados em redes neurais artificiais e lógica Fuzzy) têm recebido bastante atenção na literatura (Shah and Engell, 2011;Kramer and Morgado-Dias, 2018). Não…”

Section: Introductionunclassified

Estudo Comparativo do Projeto de Controladores PI Multivariáveis baseados em Normas de Sinais e Sistemas

Gonçalves

Negreiros

2019

Anais Do 14º Simpósio Brasileiro De Automação Inteligente

View full text Add to dashboard Cite

Multivariable PI controllers for systems with time delays can be tuned with applied optimization using different formulations for the objective function. Strategies based on H ∞ and H 2 system norms impose the need of approximation of the time delays by Padé functions. On the other hand, optimization of a performance index based on signal norms rather than system norms are more direct and straightforward to formulate, although more time-consuming. In this paper, it is analysed how the different ways to formulate the optimization problem in terms of system and signal norms can impact on the closed loop performance and on computational effort. A centralized PI controller is tuned with Differential Evolution method using a suite of different control cost functions. Simulations are performed on a 2 × 2 distillation column and the results compared. Resumo: Controladores PI multivariáveis para sistemas com atrasos de transporte podem ser sintonizados por meio de otimização usando-se diferentes formulações da função objetivo. Estratégias baseadas em normas H ∞ e H 2 colocam a necessidade de aproximar os atrasos de transporte com funções de Padé. Por outro lado, a otimização de umíndice de desempenho baseado em normas de sinais ao invés de normas de sistemasé mais direta e simples de ser formulada, embora o custo computacional seja maior. Neste trabalhoé analisado como as diferentes maneiras de formular o problema de otimização em termos de normas de sistemas e de sinais impacta no desempenho do sistema em malha fechada e no esforço computacional. Um controlador PI centralizadoé sintonizado usando o método Evolução Diferencial e uma suíte de diferentes funções de custo. Simulações são executadas num modelo de coluna de destilação 2 × 2 e os resultados são comparados.

show abstract

Applications of Artificial Neural Networks in Process Control Applications: A Review

Cited by 12 publications

References 19 publications

Energy and Production Efficiency Optimization of an Ethylene Plant Considering Process Operation and Structure

Energy and Production Efficiency Optimization of an Ethylene Plant Considering Process Operation and Structure

Comparison of ReLU and linear saturated activation functions in neural network for universal approximation

Estudo Comparativo do Projeto de Controladores PI Multivariáveis baseados em Normas de Sinais e Sistemas

Contact Info

Product

Resources

About