• Steel and Composite Structures
• /
• v.39 no.3
• /
• pp.319-335
• /
• 2021

This study proposes a new and highly-accurate artificial intelligence model, namely ANN-IP, which combines an interior-point (IP) algorithm and artificial neural network (ANN), to improve the axial compression capacity prediction of elliptical concrete-filled steel tubular (CFST) columns. For this purpose, 145 tests of elliptical CFST columns extracted from the literature are used to develop the ANN-IP model. In this regard, axial compression capacity is considered as a function of the column length, the major axis diameter, the minor axis diameter, the thickness of the steel tube, the yield strength of the steel tube, and the compressive strength of concrete. The performance of the ANN-IP model is compared with the ANN-LM model, which uses the robust Levenberg-Marquardt (LM) algorithm to train the ANN model. The comparative results show that the ANN-IP model obtains more magnificent precision (R² = 0.983, RMSE = 59.963 kN, a20 - index = 0.979) than the ANN-LM model (R² = 0.938, RMSE = 116.634 kN, a20 - index = 0.890). Finally, a new Graphical User Interface (GUI) tool is developed to use the ANN-IP model for the practical design. In conclusion, this study reveals that the proposed ANN-IP model can properly predict the axial compression capacity of elliptical CFST columns and eliminate the need for conducting costly experiments to some extent.

• Structural Engineering and Mechanics
• /
• v.70 no.6
• /
• pp.671-681
• /
• 2019

This paper examines the applicability of artificial neural network (ANN) and multivariate adaptive regression splines (MARS) to predict the compressive strength of bacteria incorporated geopolymer concrete (GPC). The mix is composed of new bacterial strain, manufactured sand, ground granulated blast furnace slag, silica fume, metakaolin and fly ash. The concentration of sodium hydroxide (NaOH) is maintained at 8 Molar, sodium silicate ($Na_2SiO_3$) to NaOH weight ratio is 2.33 and the alkaline liquid to binder ratio of 0.35 and ambient curing temperature ($28^{\circ}C$) is maintained for all the mixtures. In ANN, back-propagation training technique was employed for updating the weights of each layer based on the error in the network output. Levenberg-Marquardt algorithm was used for feed-forward back-propagation. MARS model was developed by establishing a relationship between a set of predictors and dependent variables. MARS is based on a divide and conquers strategy partitioning the training data sets into separate regions; each gets its own regression line. Six models based on ANN and MARS were developed to predict the compressive strength of bacteria incorporated GPC for 1, 3, 7, 28, 56 and 90 days. About 70% of the total 84 data sets obtained from experiments were used for development of the models and remaining 30% data was utilized for testing. From the study, it is observed that the predicted values from the models are found to be in good agreement with the corresponding experimental values and the developed models are robust and reliable.

• Journal of the Architectural Institute of Korea Structure & Construction
• /
• v.34 no.11
• /
• pp.73-79
• /
• 2018

In this study, for the prediction of energy consumption in the ventilator, one of the components of the air conditioning system, the predicted results were analyzed and accurate by the change in the number of neurons and inputs. The input variables of the prediction model for the energy volume of the fan were the supply air flow rate, the exhaust air flow rate, and the output value was the energy consumption of the fan. A predictive model has been developed to study with the Levenbarg-Marquardt algorithm through 8760 sets of one-minute resolution. Comparison of actual energy use and forecast results showed a margin of error of less than 1% in all cases and utilization time of less than 3% with very high predictability. MBE was distributed with a learning period of 1.7% to 2.95% and a service period of 2.26% to 4.48% respectively, and the distribution rate of ${\pm}10%$ indicated by ASHRAE Guidelines 14 was high.8.

• Wind and Structures
• /
• v.36 no.6
• /
• pp.423-434
• /
• 2023

Aerodynamic force coefficients are generally obtained from traditional wind tunnel tests or computational fluid dynamics (CFD). Unfortunately, the techniques mentioned above can sometimes be cumbersome because of the cost involved, such as the computational cost and the use of heavy equipment, to name only two examples. This study proposed to build a deep neural network model to predict the aerodynamic force coefficients based on data collected from CFD simulations to overcome these drawbacks. Therefore, a series of CFD simulations were conducted using different geometric parameters to obtain the aerodynamic force coefficients, validated with wind tunnel tests. The results obtained from CFD simulations were used to create a dataset to train a multilayer perceptron artificial neural network (ANN) model. The models were obtained using three optimization algorithms: scaled conjugate gradient (SCG), Bayesian regularization (BR), and Levenberg-Marquardt algorithms (LM). Furthermore, the performance of each neural network was verified using two performance metrics, including the mean square error and the R-squared coefficient of determination. Finally, the ANN model proved to be highly accurate in predicting the force coefficients of similar bridge sections, thus circumventing the computational burden associated with CFD simulation and the cost of traditional wind tunnel tests.

• Geomechanics and Engineering
• /
• v.33 no.6
• /
• pp.567-581
• /
• 2023

The prediction and achievement of a proper rock fragmentation size is the main challenge of blasting operations in surface mines. This is because an optimum size distribution can optimize the overall mine/plant economics. To this end, this study attempts to develop four improved artificial intelligence models to predict rock fragmentation through cascaded forward neural network (CFNN) and radial basis function neural network (RBFNN) models. In this regards, the CFNN was trained by the Levenberg-Marquardt algorithm (LMA) and Conjugate gradient backpropagation (CGP). Further, the RBFNN was optimized by the Dragonfly Algorithm (DA) and teaching-learning-based optimization (TLBO). For developing the models, the database required was collected from the Midouk copper mine, Iran. After modeling, the statistical functions were computed to check the accuracy of the models, and the root mean square errors (RMSEs) of CFNN-LMA, CFNN-CGP, RBFNN-DA, and RBFNN-TLBO were obtained as 1.0656, 1.9698, 2.2235, and 1.6216, respectively. Accordingly, CFNN-LMA, with the lowest RMSE, was determined as the model with the best prediction results among the four examined in this study.

• Smart Structures and Systems
• /
• v.32 no.1
• /
• pp.37-47
• /
• 2023

Indirect determination of pan evaporation (PE) has been highly regarded, due to the advantages of intelligent models employed for this objective. This work pursues improving the reliability of a popular intelligent model, namely multi-layer perceptron (MLP) through surmounting its computational knots. Available climatic data of Fresno weather station (California, USA) is used for this study. In the first step, testing several most common trainers of the MLP revealed the superiority of the Levenberg-Marquardt (LM) algorithm. It, therefore, is considered as the classical training approach. Next, the optimum configurations of two metaheuristic algorithms, namely cuttlefish optimization algorithm (CFOA) and teaching-learning-based optimization (TLBO) are incorporated to optimally train the MLP. In these two models, the LM is replaced with metaheuristic strategies. Overall, the results demonstrated the high competency of the MLP (correlations above 0.997) in the presence of all three strategies. It was also observed that the TLBO enhances the learning and prediction accuracy of the classical MLP (by nearly 7.7% and 9.2%, respectively), while the CFOA performed weaker than LM. Moreover, a comparison between the efficiency of the used metaheuristic optimizers showed that the TLBO is a more time-effective technique for predicting the PE. Hence, it can serve as a promising approach for indirect PE analysis.

• Transactions of Materials Processing
• /
• v.33 no.2
• /
• pp.96-102
• /
• 2024

This study addresses evaluation of performance of hardening model for a titanium alloy (Ti6Al4V) based on the artificial neural network (ANN) regarding the strain rate and the temperature. Uniaxial compression tests were carried out at different strain rates from 0.001 /s to 10 /s and temperatures from 575 ℃ To 975 ℃. Using the experimental data, ANN models were trained and tested with different hyperparameters, such as size of hidden layer and optimizer. The input features were determined with the equivalent plastic strain, strain rate, and temperature while the output value was set to the equivalent stress. When the number of data is sufficient with a smooth tendency, both the Bayesian regulation (BR) and the Levenberg-Marquardt (LM) show good performance to predict the flow behavior. However, only BR algorithm shows a predictability when the number of data is insufficient. Furthermore, a proper size of the hidden layer must be confirmed to describe the behavior with the limited number of the data.

• Journal of Biomedical Engineering Research
• /
• v.16 no.2
• /
• pp.223-230
• /
• 1995

This study investigates a combined mathematical model for the quantitative estimation of regional myocardial blood flow in experimental canine coronary artery occlusion and in patients with ischemic myocardial diseases using Rb-82 dynamic myocardial positron emission tomography. The coronary thrombosis was induced using the new catheter technique by narrowing the lumen of coronary vessel gradually, which finally led to partial obstruction of coronary artery. Thirty four Rb-82 dynamic myocardial PET scans were performed sequentially for each experiment using our 5, 10 and 20 second acquisition protocol, respectively, and six to seven regions of interest were drawn on each transaxial slices, one on left ventricular chamber for input function and the others on normal and decreased perfusion myocardial segments for the flow estimation in those regions. Two compartment model and graphical analysis method have been applied to the measured sets of regional PET data, and the rate constants of influx to myocardial tissue were calculated for regional myocardial flow estimates with the two parameter fits of raw data by the Levenberg-Marquardt method. The results showed that, (I) two compartment model suggested by Kety-Schmidt, with proper modification of the measured data and volume of distribution, could be used for the simple estimation of regional myocardial blood flow, (2) the calculated regional myocardial blood flow estimates were dependent on the selection of input function, which reflected partial volume effect and left ventricular wall motion in previously used graphical analysis, and (3) mathematically fitted input and tissue time activity curves were more suitable than the direct application of the measured data in terms of convergence.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.20 no.3
• /
• pp.247-254
• /
• 2008

The double-peak normal distribution function (DPDF) suggested by Cho et al.(2004) has the problems that the extremely high and low tidal elevations are frequently generated in the Monte-Carlo simulation processes because the upper and lower limits of the DPDF are unbounded in spite of the excellent goodness-offit results. In this study, the modified DPDF is suggested by introducing the upper and lower value parameters and re-scale parameters in order to remove these problems. These new parameters of the DPDF are optimally estimated by the non-linear optimization problem solver using the Levenberg-Marquardt scheme. This modified DPDF can remove completely the unrealistically generated tidal levations and give a slightly better fit than the existing DRDF. Based on the DPDF's characteristic power, the over- and under estimation problems of the design factors are also automatically intercepted, too.

• International Journal of Concrete Structures and Materials
• /
• v.7 no.3
• /
• pp.225-238
• /
• 2013

In this study, a two-layer feed-forward neural network was constructed and applied to determine a mapping associating mix design and testing factors of cement-nano silica (NS)-rice husk ash ternary blended concrete samples with their performance in conductance to the water absorption properties. To generate data for the neural network model (NNM), a total of 174 field cores from 58 different mixes at three ages were tested in the laboratory for each of percentage, velocity and coefficient of water absorption and mix volumetric properties. The significant factors (six items) that affect the permeability properties of ternary blended concrete were identified by experimental studies which were: (1) percentage of cement; (2) content of rice husk ash; (3) percentage of 15 nm of $SiO_2$ particles; (4) content of NS particles with average size of 80 nm; (5) effect of curing medium and (6) curing time. The mentioned significant factors were then used to define the domain of a neural network which was trained based on the Levenberg-Marquardt back propagation algorithm using Matlab software. Excellent agreement was observed between simulation and laboratory data. It is believed that the novel developed NNM with three outputs will be a useful tool in the study of the permeability properties of ternary blended concrete and its maintenance.