Silvia Curteanu

Multi-objective optimization of a stacked neural network using an evolutionary hyper-heuristic

The present paper deals with the development and optimization of a stacked neural network (SNN) through an evolutionary hyper-heuristic, called NSGA-II-QNSNN. The proposed hyper-heuristic is based on the NSGA-II (Non-dominated Sorting Genetic Algorithm - II) multi-objective optimization evolutionary algorithm which incorporates the Quasi-Newton (QN) optimization algorithm. QN is used for training each neural network from the stack. The final global optimal solution provided by NSGA-II-QNSNN algorithm is a Pareto optimal front. It represents all the equally good compromises that can be made between the structural complexity of the stacked neural network and its modelling performance. The set of decision variables, which led to obtaining the set of points in the Pareto optimal front, represents the optimum values for the parameters of the stacked neural network: the number of networks in the stack, the weights for every output of the composing networks, and the number of hidden neurons in each individual neural network. Each stacked neural network determined through the optimization process was trained and tested by applying it to a real world problem: the modelling of the polyacrylamide-based multicomponent hydrogels synthesis. The neural modelling established the influence of the reaction conditions on the reaction yield and the swelling degree. The results provided by NSGA-II-QNSNN were superior, not only in terms of performance, but also in terms of structural complexity, to those obtained in our previous works, where individual or aggregated neural networks were used, but the stacks were developed manually, based on successive trials.

Ten steps modeling of electrolysis processes by using neural networks

Neural networks have been developed to model the electrolysis of wastes polluted with phenolic compounds, including phenol, 4-chlorophenol, 2,4-dichlorophenol, 2,4,6-trichlorophenol, 4-nitrophenol and 2,4-dinitrophenol. They enable the prediction the Chemical Oxygen Demand of a treated waste as a function of the initial characteristics (pollutant concentration, pH), operation conditions (temperature, current density) and current charge passed. A consistent set of experimental data was obtained by electrochemical oxidation with conductive diamond electrodes, used to treat synthetic aqueous wastes. Several modeling strategies based on simple and stacked neural networks, with different transfer functions into the hidden and output layers, have been considered to obtain a good accuracy of the model. Global errors during the training stage were under 3% and those of the validation stage were under 4%, demonstrating that the neural network based technique is appropriate for modeling the system. The generalization capability of the neural networks was also tested in realistic conditions where Chemical Oxygen Demand was predicted with errors around 5%. Therefore, the developed neural models can be used in industry to determine the required treatment period, to obtain the discharge limits in batch electrolysis processes, and it is a first step in the development of process control strategies. The ten step methodology was applied to the neural network based process modeling.

Covalent and ionic co‐cross‐linking—An original way to prepare chitosan–gelatin hydrogels for biomedical applications

The first goal of this work was to develop a method for obtaining interpenetrating gelatin (G)-chitosan (CS) networks prepared by double cross-linking (covalent followed by ionic) that exhibit hydrogel character. The second goal was to modulate their properties as a function of the preparation parameters by using neural network models. This study was therefore carried out by experiment and simulation. The covalent cross-linking resulted from the reaction between the carbonyl groups of glutaraldehyde with amino groups belonging to both polymers; the ionic cross-linking is based on the interaction between tripolyphosphate anions and protonated amine groups (ammonium ions) of the polymers. The total cross-linking density (indirectly assessed by estimating the water swelling capacity) and the ability to include hydrosoluble bioactive principles are influenced by the following process parameters: the CS/G ratio, the amount of ionic cross-linker, and the ionic cross-linking time. The prepared hydrogels were characterized with respect to their structural, morphological, and some physical properties. The hydrogels ability to load high amounts of water-soluble drugs indicates their potential use as carriers for biologically active principles in the human body. A neural network methodology was applied to model the swelling degree and caffeine loading/release capacity depending on reaction conditions; in addition, applying this method, the optimal preparation conditions have been determined, targeting pre-established values for swelling degree or maximum caffeine value. The accuracy of the results obtained through this technique proves that the neural networks are suitable tools for modeling cross-linking processes taking place complex nonlinear polymers.

Forecasting municipal solid waste generation using prognostic tools and regression analysis

For an adequate planning of waste management systems the accurate forecast of waste generation is an essential step, since various factors can affect waste trends. The application of predictive and prognosis models are useful tools, as reliable support for decision making processes. In this paper some indicators such as: number of residents, population age, urban life expectancy, total municipal solid waste were used as input variables in prognostic models in order to predict the amount of solid waste fractions. We applied Waste Prognostic Tool, regression analysis and time series analysis to forecast municipal solid waste generation and composition by considering the Iasi Romania case study. Regression equations were determined for six solid waste fractions (paper, plastic, metal, glass, biodegradable and other waste). Accuracy Measures were calculated and the results showed that S-curve trend model is the most suitable for municipal solid waste (MSW) prediction.

Optimization methodology based on neural networks and self-adaptive differential evolution algorithm applied to an aerobic fermentation process

The determination of the optimal neural network topology is an important aspect when using neural models. Due to the lack of consistent rules, this is a difficult problem, which is solved in this paper using an evolutionary algorithm namely Differential Evolution. An improved, simple, and flexible self-adaptive variant of Differential Evolution algorithm is proposed and tested. The algorithm included two initialization strategies (normal distribution and normal distribution combined with the opposition based principle) and a modified mutation principle. Because the methodology contains new elements, a specific name has been assigned, SADE-NN-1. In order to determine the most influential inputs of the models, a sensitivity analysis was applied. The case study considered in this work refer to the oxygen mass transfer coefficient in stirred bioreactors in the presence of n-dodecane as oxygen vector. The oxygen transfer in the fermentation broths has a significant influence on the growth of cultivated microorganism, the accurate modeling of this process being an important problem that has to be solved in order to optimize the aerobic fermentation process. The neural networks predicted the mass transfer coefficients with high accuracy, which indicates that the proposed methodology had a good performance. The same methodology, with a few modifications, and with the best neural network models, was used for determining the optimal conditions for which the mass transfer coefficient is maximized. A short review of the differential evolution methodology is realized in the first part of this article, presenting the main characteristics and variants, with advantages and disadvantages, and fitting in the modifications proposed within the existing directions of research.

Experimental analysis and mathematical prediction of Cd(II) removal by biosorption using support vector machines and genetic algorithms

We investigated the bioremoval of Cd(II) in batch mode, using dead and living biomass of Trichoderma viride. Kinetic studies revealed three distinct stages of the biosorption process. The pseudo-second order model and the Langmuir model described well the kinetics and equilibrium of the biosorption process, with a determination coefficient, R(2)>0.99. The value of the mean free energy of adsorption, E, is less than 16 kJ/mol at 25 °C, suggesting that, at low temperature, the dominant process involved in Cd(II) biosorption by dead T. viride is the chemical ion-exchange. With the temperature increasing to 40-50 °C, E values are above 16 kJ/mol, showing that the particle diffusion mechanism could play an important role in Cd(II) biosorption. The studies on T. viride growth in Cd(II) solutions and its bioaccumulation performance showed that the living biomass was able to bioaccumulate 100% Cd(II) from a 50 mg/L solution at pH 6.0. The influence of pH, biomass dosage, metal concentration, contact time and temperature on the bioremoval efficiency was evaluated to further assess the biosorption capability of the dead biosorbent. These complex influences were correlated by means of a modeling procedure consisting in data driven approach in which the principles of artificial intelligence were applied with the help of support vector machines (SVM), combined with genetic algorithms (GA). According to our data, the optimal working conditions for the removal of 98.91% Cd(II) by T. viride were found for an aqueous solution containing 26.11 mg/L Cd(II) as follows: pH 6.0, contact time of 3833 min, 8 g/L biosorbent, temperature 46.5 °C. The complete characterization of bioremoval parameters indicates that T. viride is an excellent material to treat wastewater containing low concentrations of metal.

Artificial neural networks: applications in chemical engineering

Abstract Artificial neural networks (ANN) provide a range of powerful new techniques for solving problems in sensor data analysis, fault detection, process identification, and control and have been used in a diverse range of chemical engineering applications. This paper aims to provide a comprehensive review of various ANN applications within the field of chemical engineering (CE). It deals with the significant aspects of ANN (architecture, methods of developing and training, and modeling strategies) in correlation with various types of applications. A systematic classification scheme is also presented, which uncovers, classifies, and interprets the existing works related to the ANN methodologies and applications within the CE domain. Based on this scheme, 717 scholarly papers from 169 journals are categorized into specific application areas and general (other) applications, including the following topics: petrochemicals, oil and gas industry, biotechnology, cellular industry, environment, health and safety, fuel and energy, mineral industry, nanotechnology, pharmaceutical industry, and polymer industry. It is hoped that this paper will serve as a comprehensive state-of-the-art reference for chemical engineers besides highlighting the potential applications of ANN in CE-related problems and consequently enhancing the future ANN research in CE field.

Modeling of oxygen mass transfer in the presence of oxygen-vectors using neural networks developed by differential evolution algorithm

The search capabilities of the Differential Evolution (DE) algorithm - a global optimization technique - make it suitable for finding both the architecture and the best internal parameters of a neural network, usually determined by the training phase. In this paper, two variants of the DE algorithm (classical DE and self-adaptive mechanism) were used to obtain the best neural networks in two distinct cases: for prediction and classification problems. Oxygen mass transfer in stirred bioreactors is modeled with neural networks developed with the DE algorithm, based on the consideration that the oxygen constitutes one of the decisive factors of cultivated microorganism growth and can play an important role in the scale-up and economy of aerobic biosynthesis systems. The coefficient of mass transfer oxygen is related to the viscosity, superficial speed of air, specific power, and oxygen-vector volumetric fraction (being predicted as function of these parameters) using stacked neural networks. On the other hand, simple neural networks are designed with DE in order to classify the values of the mass transfer coefficient oxygen into different classes. Satisfactory results are obtained in both cases, proving that the neural network based modeling is an appropriate technique and the DE algorithm is able to lead to the near-optimal neural network topology.

An elitist non-dominated sorting genetic algorithm enhanced with a neural network applied to the multi-objective optimization of a polysiloxane synthesis process

This paper presents an original software implementation of the elitist non-dominated sorting genetic algorithm (NSGA-II) applied and adapted to the multi-objective optimization of a polysiloxane synthesis process. An optimized feed-forward neural network, modeling the variation in time of the main parameters of the process, was used to calculate the vectorial objective function of NSGA-II, as an enhancement to the multi-objective optimization procedure. An original technique was utilized in order to find the most appropriate parameters for maximizing the performance of NSGA-II. The algorithm provided the optimum reaction conditions (reaction temperature, reaction time, amount of catalyst, and amount of co-catalyst), which maximize the reaction conversion and minimize the difference between the obtained viscometric molecular weight and the desired molecular weight. The algorithm has proven to be able to find the entire non-dominated Pareto front and to quickly evolve optimal solutions as an acceptable compromise between objectives competing with each other. The use of the neural network makes it also suitable to the multi-objective optimization of processes for which the amount of knowledge is limited.

On the Use of Artificial Neural Networks to Monitor a Pharmaceutical Freeze-Drying Process

This paper is focused on the use of artificial neural networks to monitor a pharmaceutical freeze-drying process. A detailed phenomenological model of the process is used to provide the dataset for the learning phase of the neural network development. Then, a methodology based on a self-adaptive differential evolution scheme is combined with a back-propagation algorithm, as local search method, for the simultaneous structural and parametric optimization of the neural network which models the freeze-drying process. Using some experimentally available measurements at a generic time t, the neural network is able to estimate the temperature of the product and the thickness of the dried cake (the amount of residual ice, as well as the sublimation flux, can be easily calculated from the cake thickness) at a future time t + Δt, for the given operating conditions (the temperature of the heating shelf and the pressure in the drying chamber). Also, the duration of the primary drying phase and the maximum product temperature in the future are predicted, in case the operating conditions are not modified. In this way it is possible to understand if it is necessary to modify the operating conditions, in case the product temperature should trespass the limit value before the ending point of the primary drying. Despite the fact that the artificial neural network is obtained using a learning set determined for specific values of heat transfer coefficient (between the heating shelf and the product at the bottom of the container) and of mass transfer resistance (of the dried cake to vapor flow), reliable and accurate estimations are also obtained in case the sensor is used to monitor a process characterized by different values of heat and mass transfer coefficients.