Hybrid Artificial Neural Network-genetic Algorithm Research Articles

An investigation of machinability in micro-EDM drilling of ZrB2/Fly ash/AA7075 hybrid aluminium metal matrix composite

This work investigates the importance of micro-drilling in Hybrid aluminum metal matrix composites (HAMMCs), particularly for the aerospace and automotive industries. The HAMMCs (AA7075/5% ZrB2/2% fly) ash was fabricated using the ultrasonic assisted stir casting method. Taguchi’s design of experiments is employed to optimize the material removal rate (MRR) and tool wear rate (TWR). Analysis of variance (ANOVA) is then used to determine the significance of the model. The study also utilizes a hybrid artificial neural network-genetic algorithm (ANN-GA) approach to optimize both MRR and TWR. By optimizing machining processes for HAMMCs, manufacturers can achieve high-precision micro holes. The optimal input parameters were found to be a feed rate of 10 μm sec−1, a capacitance of 1 nF, and a voltage of 85 V, achieving the minimum TWR and maximum MRR.

Automatic Classification of Cancerous Masses in Digital Mammograms Using Curvelet Analysis and Hybrid Genetic Algorithm-Artificial Neural Network Model

Background: Mammography is the most fundamental and widely used method for detecting breast abnormalities. Distinguishing malignant from benign lesions requires extracting relevant information, which can be challenging and time-consuming for radiologists. Computer-aided diagnosis (CAD) techniques can serve as complementary diagnostic tools, assisting radiologists in the early detection and analysis of abnormalities in mammograms. Objectives: This study aimed to propose a CAD system for extracting significant features of abnormalities in breast mammograms using Curvelet transform and fractal analysis, and classifying breast tumors as malignant or benign based on the calculated features. Patients and Methods: In this study, an efficient feature extraction method was applied, utilizing Curvelet transform and fractal analysis, on a dataset comprising 113 abnormal images from the Mammographic Image Analysis Society (MIAS) database, which included 62 benign and 51 malignant cases. The method yielded 575 features, but due to potential irrelevance or redundancy, a multi-objective optimization (MOO) approach based on a genetic algorithm (GA) for an artificial neural network (ANN), named GA-MOO-ANN, was proposed to obtain and focus on an optimal and effective feature set. As a result of this approach, a set of 17 efficient features was extracted. The proposed algorithm was implemented in MATLAB 2014a, and the performance metrics were calculated using 6-fold cross-validation. Results: The experimental results demonstrated exceptional performance, with an accuracy (Acc) of 98.2%, specificity (Sp) of 100%, sensitivity (Se) of 96.8%, positive predictive value (PPV) of 100%, negative predictive value (NPV) of 96.2%, and an impressive area under the curve (AUC) of 0.98, providing comparable results to other recent methods. Conclusion: The current findings suggest that the proposed method could be a valuable tool for breast cancer diagnosis, potentially reducing the number of unnecessary breast biopsies. This method may lead to more efficient patient evaluation and earlier detection of breast tumors.

Controlling recast layer and material removal in reverse µEDM for sustainable fabrication of microelectrode arrays

Microelectrode arrays (MEAs) are essential for researching neurons and heart tissue because they record and stimulate neuronal activity. For manufacturing MEAs on conductive materials, reverse microelectrical discharge machining, or reverse µEDM, presents a viable method. However, achieving precise MEAs requires control over recast layers generated during micromachining. This study focuses on recast layer thickness (RLT) and material removal rate (MRR) as key performance indicators for sustainable reverse µEDM fabrication of MEAs. RLT affects surface stability, while MRR influences energy consumption and cost. Optimising these responses depends on three input factors: feed rate, capacitance, and voltage. ANOVA analysis reveals that capacitance is crucial in both MRR and recast layer thickness (RLT) in reverse µEDM, contributing to 88.71% and 94.04% simultaneously. A hybrid artificial neural network-genetic algorithm for global optimisation was employed. The optimised parameters 46.5nF capacitance, 110 V voltage, and 5 µm/sec feed yielded a minimum RLT of 1.58 µm and a maximum MRR of 0.3 mm³/min. These advancements signify significant improvements in the precision and efficiency of MEAs fabrication through reverse µEDM, contributing to its sustainable implementation.

A Comparative Study on the Estimation of Wind Speed and Wind Power Density Using Statistical Distribution Approaches and Artificial Neural Network-Based Hybrid Techniques in Çanakkale, Türkiye

In recent years, wind energy has become remarkably popular among renewable energy sources due to its low installation costs and easy maintenance. Having high energy potential is of great importance in the selection of regions where wind energy investments will be made. In this study, the wind power potential in Çanakkale Province, located in the northwest of Türkiye, is examined, and the wind speed is estimated using hourly and daily data over a one-year period. The data, including 12 different meteorological parameters, were taken from the Turkish State Meteorological Service. The two-parameter Weibull and Rayleigh distributions, which are the most widely preferred models in wind energy studies, are employed to estimate the wind power potential using hourly wind speed data. The graphical method is implemented to calculate the shape (k) and scale (c) parameters of the Weibull distribution function. Daily average wind speed estimation is performed with artificial neural network–genetic algorithm (ANN-GA) and ANN–particle swarm optimization (ANN-PSO) hybrid approaches. The proposed hybrid ANN-GA and ANN-PSO algorithms provide correlation coefficient values of 0.94839 and 0.94042, respectively, indicating that the predicted and measured wind speed values are notably close. Statistical error indices reveal that the ANN-GA model outperforms the ANN-PSO model.

Hydroisomerisation and Hydrocracking of n-Heptane: Modelling and Optimisation Using a Hybrid Artificial Neural Network–Genetic Algorithm (ANN–GA)

In this paper, the focus is on upgrading the value of naphtha compounds represented by n-heptane (n-C7H16) with zero octane number using a commercial zeolite catalyst consisting of a mixture of 75% HY and 25% HZSM-5 loaded with different amounts, 0.25 to 1 wt.%, of platinum metal. Hydrocracking and hydroisomerisation processes are experimentally and theoretically studied in the temperature range of 300–400 °C and under various contact times. A feedforward artificial neural network (FFANN) based on two hidden layers was used for the purpose of process modelling. A total of 80% of the experimental results was used to train the artificial neural network, with the remaining results being used for evaluation and testing of the network. Tan-sigmoid and log-sigmoid transfer functions were used in the first and second hidden layers, respectively. The optimum number of neurons in hidden layers was determined depending on minimising the mean absolute error (MAE). The best ANN model, represented by the multilayer FFANN, had a 4–24–24–12 topology. The ANN model accurately simulates the process in which the correlation coefficient (R2) was found to be 0.9918, 0.9492, and 0.9426 for training, validation, and testing, respectively, and an average of 0.9767 for all data. In addition, the operating conditions of the process were optimised using the genetic algorithm (GA) towards increasing the octane number of the products. MATLAB® Version 2020a was utilised to complete all required computations and predictions. Optimal operating conditions were found through the theoretical study: 0.85 wt.% Pt-metal loaded, 359.36 °C, 6.562 H2/n-heptane feed ratio, and 3.409 h−1 weight-hourly space velocity (WHSV), through which the maximum octane number (RON) of 106.84 was obtained. Finally, those operating conditions largely matched what was calculated from the results of the experimental study, where the highest percentage of the resulting isomers was found with about 78.7 mol% on the surface of the catalyst loaded with 0.75 wt.% Pt-metal at 350 °C using a feed ratio of 6.5 H2/n-C7 and WHSV of 2.98 h−1.

Optimizing Fenton-like process, homogeneous at neutral pH for ciprofloxacin degradation: Comparing RSM-CCD and ANN-GA.

Antibiotics are considered among the most non-biodegradable environmental contaminants due to their genetic resistance. Considering the importance of antibiotics removal, this study was aimed at multi-objective modeling and optimization of the Fenton-like process, homogeneous at initial circumneutral pH. Two main issues, including maximizing Ciprofloxacin (CIP) removal and minimizing sludge to iron ratio (SIR), were modeled by comparing central composite design (CCD) based on Response Surface Methodology (RSM) and hybrid Artificial Neural Network-Genetic Algorithm (ANN-GA). Results of simultaneous optimization using ethylene diamine tetraacetic acid (EDTA) revealed that at pH≅7, optimal conditions for initial CIP concentration, Fe2+ concentration, [H2O2]/[Fe2+] molar ratio, initial EDTA concentration, and reaction time were 14.9mg/L, 9.2mM, 3.2, 0.6mM, and 25min, respectively. Under these optimal conditions, CIP removal and SIR were predicted at 85.2% and 2.24 (gr/M). In the next step, multilayer perceptron (MLP) and radial basis function (RBF) artificial neural networks (ANN) were developed to model CIP and SIR. It was concluded that ANN, especially multilayer perceptron (MLP-ANN) has a decent performance in predicting response values. Additionally, multi-objective optimization of the process was performed using Genetic Algorithm (GA) and Non-dominated Sorting Genetic Algorithm-II (NSGA-II) to maximize CIP removal efficiencies while minimizing SIR. NSGA-II optimization algorithm showed a reliable performance in the interaction between conflicting goals and yielded a better result than the GA algorithm. Finally, TOPSIS method with equal weights of the criteria was applied to choose the best alternative on the Pareto optimal solutions of the NSGA-II. Comparing the optimal values obtained by the multi-objective response surface optimization models (RSM-CCD) with the NSGA-II algorithm showed that the optimal variables in both models were close and, according to the absolute relative error criterion, possessed almost the same performance in the prediction of variables.

Production of Blending Quality Bioethanol from Broken Rice: Optimization of Process Parameters and Kinetic Modeling.

The enzymatic and bio-enzymatic saccharification of waste broken rice (79.8% of starch) was successfully carried out to produce a reducible sugar solution (160g/L). Bioethanol of concentration 71.2g/L (9.0% v/v) was prepared by fermentation of reducing sugar solution, using the commercially available waste brewer's yeast (Saccharomyces cerevisiae). The fermentation process parameters were optimized through response surface methodology (RSM) and hybrid artificial neural network-genetic algorithm (ANN-GA) for optimizing the ethanol concentration. The hybrid ANN-GA model predicted a maximum concentration of 71.9g/L with a deviation of only 0.97% from the experimental value (71.2g/L). Four different kinetic models were attempted to fit the experimental time evolution of concentrations with the kinetic parameters estimated by the Levenberg-Marquardt optimization technique. The 4th order Runge-Kutta algorithm was implemented through a C program module. The accuracy of each model was checked against coefficient of determination R2, adjusted R2, the absolute mean deviation (AMD), and root mean square deviation (RMSD). The Andrew-Levenspiel kinetics produced the best performance criteria at two initial substrate concentrations of 160 and 170g/L. Finally, the FTIR analysis of 781.2g/L (98.5% v/v) bioethanol (concentrated by two-stage vacuum distillation followed by treatment with 3A molecular sieve) showed a favorable blending possibility with the commercial gasoline (petrol) as a green fuel.

Multi-objective optimisation of ultrasonically welded dissimilar joints through machine learning

The use of composite materials is increasing in industry sectors such as renewable energy generation and storage, transport (including automotive, aerospace and agri-machinery) and construction. This is a result of the various advantages of composite materials over their monolithic counterparts, such as high strength-to-weight ratio, corrosion resistance, and superior fatigue performance. However, there is a lack of detailed knowledge in relation to fusion joining techniques for composite materials. In this work, ultrasonic welding is carried out on a carbon fibre/PEKK composite material bonded to carbon fibre/epoxy composite to investigate the influence of weld process parameters on the joint’s lap shear strength (LSS), the process repeatability, and the process induced defects. A 33 parametric study is carried out and a robust machine learning model is developed using a hybrid genetic algorithm–artificial neural network (GA–ANN) trained on the experimental data. Bayesian optimisation is employed to determine the most suitable GA–ANN hyperparameters and the resulting GA–ANN surrogate model is exploited to optimise the welding process, where the process performance metrics are LSS, repeatability and joint visual quality. The prediction for the optimal LSS was subsequently validated through a further set of experiments, which resulted in a prediction error of just 3%.

Investigations on Mechanical Behaviour and Failure Mechanism of Resistance Spot-Welded DP590 Steel Using Artificial Neural Network

Modelling of the spot weld responses, i.e. shear-tensile strength, coach-peel strength, mean dynamic contact resistance and nugget diameter, was carried out along with the optimization of weld process parameters, i.e. weld current, weld time and electrode force. The experiments for weld fabrication were designed using face-centred central composite design under response surface methodology. Experimental results acquired from the designed experiments were used to generate prediction models of the responses using the artificial neural network approach. Hybrid artificial neural network-genetic algorithm technique was used in order to obtain the optimized welding process parameters. Validation checks were done for the optimized parameters and responses in order to justify the adequacy of the models and optimized value. Failure mode analysis was done to determine the plasticity fraction and to relate this with the mode of failure and microhardness values. The shear-tensile strength, coach-peel strength and nugget diameter increase with the corresponding increase in the amount of heat input during the weld fabrication. The optimum process parameters are 8.98 kA of weld current, 284.92 ms of weld time and 3.04 kN of electrode force leading to the response values as shear-tensile strength of 21.93 kN.

Process optimization of catalytic steam reforming of toluene to hydrogen using response surface methodology (RSM) and artificial neural network-genetic algorithm (ANN-GA)

Catalytic steam reforming of toluene (SRT) over nickel-cobalt supported on modified activated carbon for hydrogen production has been investigated. The center composite design of experiment in response surface methodology (RSM) was initially applied to optimize the catalytic SRT for hydrogen production before being utilized in the model building of the hybrid artificial neural network-genetic algorithm (ANN-GA). The genetic algorithm was carried out over the ANN model to achieve the maximum target response. The process optimization modeling using the best fitness function gave an insight of the optimal operating condition in SRT over the prepared catalyst. The results conferred that maximum hydrogen yield could be obtained at the optimal conditions of 700 °C temperature, 0.034 ml/min feed flow rate, 0.1 g catalyst loading and S/C ratio of 1 by ANN-GA model, and 762 °C temperature, 0.022 ml/min feed flow rate, 0.3 g catalyst loading and S/C ratio of 5.6 by the RSM model. Predicted results from ANN model were in higher agreement with the experimental data at R2=0.95 compared with the RSM model.

Multiwalled Carbon Nanotube-N-Doped Graphene/Poly(3,4-ethylenedioxythiophene):Poly(styrenesulfonate) Nanohybrid for Electrochemical Application in Intelligent Sensors and Supercapacitors.

In this study, we reported the preparation of a conducting polymeric/inorganic nanohybrid consisting of multiwalled carbon nanotubes (MWCNT), N-doped graphene (NGr), and poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS), and its electrochemical application in intelligent sensors and supercapacitors. The multilayer thin film of the PEDOT:PSS-supported MWCNT-NGr nanohybrid was prepared by a facile layer-by-layer assembly strategy. The obtained conducting polymeric/inorganic nanohybrid modified electrode displayed superior electron transfer ability and a high specific surface area, which was used for electrochemical applications in intelligent sensors and supercapacitors. Remarkably, the fabricated amaranth sensor exhibited a broad linear range of 0.05–10 μM with a limit of detection of 0.015 μM under the optimal conditions. With the help of the response surface methodology, multivariate optimization was used as a substitute for the traditional single variable optimization to reflect the complete real effects of multivariate optimization in a sensing platform. Machine learning implemented by hybrid genetic algorithm-artificial neural network was used as an intelligent analysis model to replace the traditional regression analysis model for realizing intelligent analysis and output of sensing system. The MWCNT-NGr/PEDOT:PSS modified electrode exhibited a considerable specific capacitance of 6.5 mF cm–2 at a current density of 2.0 mA cm–2. The proposed results provided a new thought for a nanosensing platform equipped with a supercapacitor as a self-powered electrochemical energy storage system and machine learning as an intelligent analysis and output system in the near future.

Prediction of concrete strength in presence of furnace slag and fly ash using Hybrid ANN-GA (Artificial Neural Network-Genetic Algorithm)

Mineral admixtures have been widely used to produce concrete. Pozzolans have been utilized as partially replacement for Portland cement or blended cement in concrete based on the materials' properties and the concrete's desired effects. Several environmental problems associated with producing cement have led to partial replacement of cement with other pozzolans. Furnace slag and fly ash are two of the pozzolans which can be appropriately used as partial replacements for cement in concrete. However, replacing cement with these materials results in significant changes in the mechanical properties of concrete, more specifically, compressive strength. This paper aims to intelligently predict the compressive strength of concretes incorporating furnace slag and fly ash as partial replacements for cement. For this purpose, a database containing 1030 data sets with nine inputs (concrete mix design and age of concrete) and one output (the compressive strength) was collected. Instead of absolute values of inputs, their proportions were used. A hybrid artificial neural network-genetic algorithm (ANN-GA) was employed as a novel approach to conducting the study. The performance of the ANN-GA model is evaluated by another artificial neural network (ANN), which was developed and tuned via a conventional backpropagation (BP) algorithm. Results showed that not only an ANN-GA model can be developed and appropriately used for the compressive strength prediction of concrete but also it can lead to superior results in comparison with an ANN-BP model.

Evaluation of material properties using planetary ball milling for modeling the change of particle size distribution in a gas-solid fluidized bed using a hybrid artificial neural network-genetic algorithm approach

This study aims to use planetary ball-milling as an evaluation tool of material properties and the result is subsequently used to develop a model for the change in particle size distribution (PSD) during fluidization for a range of materials using artificial neural network (ANN) method. It is believed that material properties such as hardness, density, brittleness, structure, etc play a crucial role in the particle attrition behavior. Unfortunately, little information on material properties is available, considering the wide variety of materials present. As a result, planetary ball-milling is proposed as a fast assessment technique to identify the properties of different materials. Planetary ball milling devices are readily available in most laboratories and the reduction in PSD can resemble the particle attrition process during fluidization. A Rosin-Rammler (RR) distribution was used to describe the PSD for both fluidization and ball milling processes.

Prediction of soil wind erodibility using a hybrid Genetic algorithm – Artificial neural network method

Understanding and consequent modeling of soil wind erodibility is hampered by the complex nature of the eroding processes and limited empirical data. It is often necessary to resort to robust approaches capable of finding correlated patterns among soil erodibility magnitudes and their drivers. To signify soil erodibility to wind, we used a portable wind tunnel to measure wind erosion rate (g m−2 s−1) at a total of 118 sites in Kerman Province, southeast Iran. At each sampling site, 17 different factors affecting soil erodibility were measured. Gravel coverage, surface crust, very fine and very coarse sands, aggregate stability, and calcium carbonate equivalent (CCE) were introduced as the more important parameters affecting soil erodibility by hybrid Genetic Algorithm-Artificial Neural Network (GA-ANN). A Multi-Layer Perception (MLP) neural network was developed to predict erodibility changes in response to spatial variation of the selected features. The developed MLP-model provided a strong basis for the prediction of soil erodibility, where the coefficient of determination (R2) values of 0.89 and 0.87 were obtained by comparing the measured and predicted wind erosion rates for the training and testing data, respectively. The acceptable levels of the statistical validation criteria were also an indication of the proper performance of the model. Furthermore, the soil erodibility was sensitive respectively to surface crust, very fine sand, and very coarse sand parameters.

Estimation of near-saturated soil hydraulic properties using hybrid genetic algorithm-artificial neural network

Near-saturated hydraulic properties are the key parameters for water transport models in the unsaturated zone and essential for management practices. This study was conducted to compare efficacy of multiple linear regression (MLR) and hybrid method of genetic algorithm with artificial neural network (GA-ANN) for prediction of near-saturated soil hydraulic properties in Moghan plain, north-western Iran. The results of MLR analysis indicated that this method had low potential to predict near-saturated soil hydraulic properties in the study area, which only could explain 14–38% of variability in the studied properties. Otherwise, GA-ANN was much higher powerful that could explain about 35–80% of total variability in the mentioned properties in the study area. The results of sensitivity analysis suggest that soil particle size distribution, organic matter, electrical conductivity and relative bulk density were the most crucial with variety of priorities for explaining variability of the near-saturated soil hydraulic properties in the study area in the semiarid region. In overall, it was concluded that application of intelligent system using the easily available soil properties as predictors could provide reliable estimates of near-saturated soil hydraulic properties at the filed scale.

Modeling azo dye removal by sono-fenton processes using response surface methodology and artificial neural network approaches

Textile industry wastewaters, which cause serious problems in the environment and human health, include synthetic dyes, complex organic pollutants, surfactants, and other toxic chemicals and therefore must be removed by advanced treatment methods. Determination of appropriate treatment conditions for efficient use of advanced treatment methods is an important and necessary step. In the last thirty years, the Artificial Neural Network-Genetic Algorithm (ANN-GA) and Response Surface Methodology (RSM) have emerged as the most effective empirical modeling and optimization methods especially for nonlinear systems. Reactive Red 195 azo dyestuff was chosen as the target pollutant. The color removal efficiency was modeled and optimized as a function of Sono-Fenton conditions such as H2O2 dosage, Fe2+ dosage, initial pH value, ultrasound power, and ultrasound frequency, using ANN-GA and RSM. The generalization and predictive ability of these methods were compared using the results of the 46 experimental sets generated by the Box-Behnken design. The mean square errors for these models are 3.01612 and 0.00295, and the regression coefficients showing the superiority of ANN in determining nonlinear behavior are 0.9856 and 0.9164, respectively. In optimal conditions, the prediction errors with hybrid ANN-GA and RSM models are 0.002% and 3.225%, respectively.

Hybrid neural network–particle swarm optimization algorithm and neural network–genetic algorithm for the optimization of quality characteristics during CO2 laser cutting of aluminium alloy

Aerospace and automobile industries employ aluminium alloys due to its lightweight and excellent resistance to corrosion. As aluminium is highly reflective and highly thermally conductive, it is difficult-to-cut material by laser processing. The quality characteristics of the cut predominantly depend upon the combination of laser processing parameters. The main quality indices for evaluating CO2 laser cutting were surface roughness, kerf width and kerf taper; and the machining parameters considered were laser power, speed and gas pressure. This work suggests hybrid artificial neural network (ANN)–particle swarm optimization (PSO) algorithm and artificial neural network (ANN)–genetic algorithm (GA) to optimize the associated multi-response characteristics during CO2 laser cutting of aluminium 6061 alloys. The results illustrate that the hybrid ANN–GA and ANN–PSO model is an efficient tool for the optimization of process parameters in CO2 laser cutting of difficult-to-cut material—aluminium. From the optimization results, it can be concluded that the proposed ANN–GA approach can be efficiently utilized to optimize the parameters for obtaining minimum roughness, kerf width and kerf taper.

GA-BP in Thermal Fatigue Failure Prediction of Microelectronic Chips

A thermal fatigue life prediction model of microelectronic chips based on thermal fatigue tests and solder/substrate interfacial singularity analysis from finite element method (FEM) analysis is established in this paper. To save the calculation of interfacial singular parameters of new chips for life prediction, and improve the accuracy of prediction results in actual applications, a hybrid genetic algorithm–artificial neural network (GA–ANN) strategy is utilized. The proposed algorithm combines the local searching ability of the gradient-based back propagation (BP) strategy with the global searching ability of a genetic algorithm. A series of combinations of the dimensions and thermal mechanical properties of the solder and the corresponding singularity parameters at the failure interface are used to train the proposed GA-BP network. The results of the network, together with the established life prediction model, are used to predict the thermal fatigue lives of new chips. The comparison between the network results and thermal fatigue lives recorded in experiments shows that the GA-BP strategy is a successful prediction technique.

Effects of soil properties, water quality and management practices on pistachio yield in Rafsanjan region, southeast of Iran

In recent decades, the quantity and quality of irrigation water have been reduced due to a significant increase in pistachio cultivation and uncontrolled exploitation of groundwater resources as well as reduction in rainfall precipitation. Therefore, agricultural producers, researchers and policy makers need to pay more attention to appropriate land management as an important strategy to achieve greater yield per unit area and to use soil and water resources in an optimal way. So, the present study was conducted to model the relationships between pistachio yield and soil, water and management variables in Rafsanjan region, the southeast of Iran. One hundred and ninety nine mature orchards were selected and sampled in such a way that an acceptable range of soil and water quality and management were included. The data set consisted of a dependent variable (pistachio yield) and 67 independent variables including soil, water and management characteristics. The results of hybrid genetic algorithm-artificial neural network (GA-ANN) showed that the lowest error was related to the case in which the 23 features were used in modeling. Then, stepwise multiple linear regression (MLR) and artificial neural network (ANN) techniques were applied to estimate pistachio yield. The results indicated that MLR could explain only 28% of the pistachio yield variation, whereas its accuracy increased when the data became more homogeneous via geographically dividing the study area into four parts with the highest densities of pistachio orchards. ANN-based model had a 90% accuracy to predict pistachio yield in the study area. Thus, an accurate estimation of pistachio yield could be achieved by reducing the data dimensionality using feature selection techniques and modeling with ANN. As the models were highly sensitive to irrigation water features, special attention should be paid to new irrigation methods and management practices as an effective strategy to minimize water losses and increase water use efficiency.

Artificial neural Network−Genetic algorithm modeling for moisture content prediction of savory leaves drying process in different drying conditions

In this study, the application of a versatile approach for modeling and prediction of the moisture content of dried savory leaves using hybrid artificial neural network-genetic algorithm has been presented. Genetic Algorithm was used in order to find the best Feed Forward Neural Network (FFNN) structure for modeling and estimation of moisture content in the drying process of savory leaves. The experiments were performed at three air temperatures of 40, 60 and 80 °C and at three levels of relative humidity 20%, 30% and 40% and air velocity of 1, 1.5 and 2.0 m/s for drying the savory leaves in the forced conductive dryer. Optimized neural network by GA had two hidden layers with 9 and 17 neurons in first and second hidden layers, respectively. Mean Square Error (MSE) value (0.000094606) and correlation coefficient (0.9992) of FFNN-GA experiments showed that moisture content can be accurately predicted from the input variables: air temperature, airflow velocity, relative humidity and drying time. Moreover, results showed that the optimized neural network topology could denote the superior ability of this intelligent model for on-line prediction of the moisture content of Savory leaves in different drying conditions.