Comparison between different methods for developing neural network topology applied to a complex polymerization process

Curteanu, Silvia; Leon, Florin; Furtuna, Renata; Drăgoi, Elena Niculina; Curteanu, Neculai

doi:10.1109/ijcnn.2010.5596592

Cited by 9 publications

(6 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Also, their focus was on the ANN with two hidden layers. 37 Additionally, coupling the ANN-GA techniques has been extensively applied for predicting and finding the optimal parameters in scientific surveys and engineering design. 36 Besides, such method has been employed for optimizing the structural features of the biological scaffolds for meeting the needed mechanical features of native tissues.…”

Section: Introductionmentioning

confidence: 99%

Modeling and optimization of the core-shell nanofibrous composite mat as a scaffold via hybrid models

Haghdoost

Razbin

Bahrami

et al. 2022

Journal of Industrial Textiles

View full text Add to dashboard Cite

It has been a challenging subject for researchers to manipulate the electrospinning key factors to achieve a composite nanofiber with proper properties. In this study, an experiment according to central composite design to investigate the effect of parameters including polyvinylpyrrolidone concentration, zeolite concentration, voltage, core flow rate and shell flow rate on diameter, maximum strength and porosity of polyethersulfone/polyvinylpyrrolidone/zeolite core-shell composite nanofiber has been designed. Later on, two sets of models consisting of response surface methodology and artificial neural network are trained. Then, their performance was evaluated based on the definition of a novel goodness function. In the next step, the genetic algorithm is used to find the optimal design for scaffold applications. The results demonstrated that the average goodness value of models based on an artificial neural network ([Formula: see text] 1.999) is higher than response surface methodology ones ([Formula: see text] 1.780). Additionally, the genetic algorithm was able to find an optimal design with lower cost value (0.006) than the optimum sample (0.113) among the produced ones. Finally, the scanning electron microscopy micrographs highlighted that there is a strong and good cell proliferation on the selected design of nanofiber composite mat as the optimum scaffold.

show abstract

Section: Introductionmentioning

confidence: 99%

Modeling and optimization of the core-shell nanofibrous composite mat as a scaffold via hybrid models

Haghdoost

Razbin

Bahrami

et al. 2022

Journal of Industrial Textiles

View full text Add to dashboard Cite

show abstract

“…The simultaneous topological and structural optimization of a neural model for the styrene polymerization was performed using three methods: a six-step optimization methodology based on a systematized trial and error applied for ANN parameters (OMP method), a classic DE variant, and a GA approach (Curteanu et al 2010). A model able to predict monomer conversion, numerical average molecular weight, and gravimetrical average molecular weight was created using initiator concentration, temperature, and reaction time as input variables.…”

Section: Artificial Neural Networkmentioning

confidence: 99%

The use of differential evolution algorithm for solving chemical engineering problems

Drăgoi

Curteanu

2016

Reviews in Chemical Engineering

Self Cite

View full text Add to dashboard Cite

AbstractDifferential evolution (DE), belonging to the evolutionary algorithm class, is a simple and powerful optimizer with great potential for solving different types of synthetic and real-life problems. Optimization is an important aspect in the chemical engineering area, especially when striving to obtain the best results with a minimum of consumed resources and a minimum of additional by-products. From the optimization point of view, DE seems to be an attractive approach for many researchers who are trying to improve existing systems or to design new ones. In this context, here, a review of the most important approaches applying different versions of DE (simple, modified, or hybridized) for solving specific chemical engineering problems is realized. Based on the idea that optimization can be performed at different levels, two distinct cases were considered – process and model optimization. In both cases, there are a multitude of problems solved, from different points of view and with various parameters, this large area of successful applications indicating the flexibility and performance of DE.

show abstract

“…The effect of the interaction between factors is also neglected, leading to a low efficiency in process optimization, which is not guaranteed to find the optimal value [ 19 ]. Works by [ 20 , 21 ] have successfully optimized the network setting such as the number of neurons, the epoch and training function using the genetic algorithm (GA), and differential evaluation (DE). There is no specific rule used in selecting the value of network parameters, and this process is dependent on the complexity of the modelled system.…”

Section: Introductionmentioning

confidence: 99%

“…It is most important to find the network parameters, which will affect the determination of weight and bias for model development. Several pieces of literature focus on the improvement of network training by optimizing the weights and biases using other optimization approaches such as GA and PSO [ 21 , 22 ]. The response surface methodology (RSM) method has successfully adopted in the literature to optimize the data from process industries, mainly for the membrane bioreactor treatment process [ 23 , 24 , 25 ].…”

Section: Introductionmentioning

confidence: 99%

Improved Artificial Neural Network Training Based on Response Surface Methodology for Membrane Flux Prediction

Ibrahim

Wahab

2022

Membranes

View full text Add to dashboard Cite

This paper presents an improved artificial neural network (ANN) training using response surface methodology (RSM) optimization for membrane flux prediction. The improved ANN utilizes the design of experiment (DoE) technique to determine the neural network parameters. The technique has the advantage of training performance, with a reduced training time and number of repetitions in achieving good model prediction for the permeate flux of palm oil mill effluent. The conventional training process is performed by the trial-and-error method, which is time consuming. In this work, Levenberg–Marquardt (lm) and gradient descent with momentum (gdm) training functions are used, the feed-forward neural network (FFNN) structure is applied to predict the permeate flux, and airflow and transmembrane pressure are the input variables. The network parameters include the number of neurons, the learning rate, the momentum, the epoch, and the training functions. To realize the effectiveness of the DoE strategy, central composite design is incorporated into neural network methodology to achieve both good model accuracy and improved training performance. The simulation results show an improvement of more than 50% of training performance, with less repetition of the training process for the RSM-based FFNN (FFNN-RSM) compared with the conventional-based FFNN (FFNN-lm and FFNN-gdm). In addition, a good accuracy of the models is achieved, with a smaller generalization error.

show abstract

Comparison between different methods for developing neural network topology applied to a complex polymerization process

Cited by 9 publications

References 30 publications

Modeling and optimization of the core-shell nanofibrous composite mat as a scaffold via hybrid models

Modeling and optimization of the core-shell nanofibrous composite mat as a scaffold via hybrid models

The use of differential evolution algorithm for solving chemical engineering problems

Improved Artificial Neural Network Training Based on Response Surface Methodology for Membrane Flux Prediction

Contact Info

Product

Resources

About