Investigating the Performance of Neural Network Based Group Method of Data Handling to Pan's Daily Evaporation Estimation (Case Study: Garmsar City)
Hojat
Karami
Faculty of Civil Engineering, Semnan University, Semnan, Iran
author
Hamidreza
Ghazvinian
Faculty of Civil Engineering, Semnan University, Semnan, Iran
author
Mohammadhassan
Dehghanipour
Faculty of Civil Engineering, Semnan University, Semnan, Iran
author
Mohammad
Ferdosian
Faculty of Civil Engineering, K. N. Toosi University of Technology, Tehran, Iran
author
text
article
2021
eng
Evaporation is a complex and nonlinear phenomenon due to the interactions of different climatic factors. Therefore, advanced models should be used to estimate evaporation. In the present study, the Neural Network-Based Group Method of Data Handling was used to estimate and simulate the evaporation rate from the pan in the synoptic station of Garmsar city located in Semnan province, Iran. For this purpose, the daily meteorological data of evaporation, minimum and maximum temperature, wind speed, relative humidity, air pressure, and sunny hours of the said station during the nine years (2009-2018) were used. The percent of data on training, test, number of the used layers, and the highest number of neurons were considered as 60%, 40%, 5%, and 30%, respectively. The studied method's accuracy was investigated using the statistical parameter of Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and correlation coefficient, and. Sensitivity analysis of the input parameters was performed using the GMDH-NN model. This study showed that R2, RMSE, and MAE values in the test phase were obtained as 0.84, 2.65, and 1.91, respectively, in the most optimal state. From the third layer onwards, the amount of the best mean squared errors of the Validation data have converged to 0.062, and it is not affordable to use more layers for the modeling of the evaporation pan in the Garmsar station. The standard deviation and mean amounts of the errors are -0.1210 and 2.552 respectively. The amounts of the best mean squared errors of the validation data are presented. It shows that although the layers are increased, the amounts of the mean squared errors have not changed considerably. (Maximum 0.003). The sensitivity analysis results showed that the two input parameters of minimum temperature and relative humidity percent have a higher effect on evaporation pan modeling than other input parameters.
Journal of Soft Computing in Civil Engineering
Pouyan Press
2588-2872
5
v.
2
no.
2021
1
18
https://www.jsoftcivil.com/article_129387_70d365537fff97c0c35f3deae549ae38.pdf
dx.doi.org/10.22115/scce.2021.274484.1282
Modelling Slump of Concrete Containing Natural Coarse Aggregate from Bida Environs Using Artificial Neural Network
Abdulazeez
Yusuf
Lecturer II, Department of Civil Engineering, Faculty of Engineering, Federal University of Technology, Minna, Nigeria
author
Mohammed
Abdullahi
Professor, Department of Civil Engineering, Federal University of Technology, Minna, Nigeria
author
Salawu
Sadiku
Professor, Department of Civil Engineering, Federal University of Technology, Minna, Nigeria
author
James
Aguwa
Professor, Department of Civil Engineering, Federal University of Technology, Minna, Nigeria
author
Bala
Alhaji
Lecturer I, Department of Civil Engineering, Federal University of Technology, Minna, Nigeria
author
Taliha
Folorunso
Lecturer I, Department of Mechatronics Engineering, Federal University of Technology, Minna, Nigeria
author
text
article
2021
eng
Consumption of crushed granite as coarse aggregate in concrete has led to devastating environmental and ecological consequences. In order to preserve local and urban ecology therefore, substitute aggregate such as naturally occurring stone with the propensity of reducing this problem was studied. Furthermore, artificial Neural Network (ANN) models have become the preferred modeling approach due to their accuracy. Thus, in this paper, MATLAB software was used to develop ANN models for predicting slump of concrete made using Bida Natural Gravel (BNG). Four model architectures (5:5:1; 5:10:1; 5:15:1 and 5:20:1) were tried using a back-propagation algorithm with a tansig activation function. The performance of the developed models was examined using Mean Square Error (MSE), Correlation Coefficient (R) and Nash-Sutcliffe Efficiency (NSE). Results showed that 5:20:1 model architecture with MSE of 8.33e-27, R value of 98% and NSE of 0.96 was the best model. The chosen 5:20:1 ANN model also out performed Multiple Linear Regression (MLR) model which recorded MSE of 0.83, R value of 88.68% and NSE of 0.87. The study concluded that the higher the neuron in hidden layer of ANN slump model for concrete containing BNG, the better the model.
Journal of Soft Computing in Civil Engineering
Pouyan Press
2588-2872
5
v.
2
no.
2021
19
38
https://www.jsoftcivil.com/article_129388_f04bb5efa726039f922d24995f3ff33f.pdf
dx.doi.org/10.22115/scce.2021.268839.1272
Application of Contourlet Transform in Damage Localization and Severity Assessment of Prestressed Concrete Slabs
Hashem
Jahangir
Assistant Professor, Department of Civil Engineering, University of Birjand, Birjand, Iran
author
Mohsen
Khatibinia
Associate Professor, Department of Civil Engineering, University of Birjand, Birjand, Iran
author
Mahdi
Kavousi
M.Sc. in Structural Engineering, Civil Engineering Department, Hormozan University of Birjand, Birjand, Iran
author
text
article
2021
eng
In this paper, the location and severity of damages in prestressed concrete slabs are assessed using the contourlet transform as a novel signal processing method. To achieve this goal, the numerical models of prestressed concrete slabs were built based on the experimental specimens reported in the previous research works. Then, the single, double, and triple damage scenarios with various geometric shapes (transverse, longitudinal, inclined, and curved slots) at different positions (middle and corners) were created in the models. To assess the severity of damages, the depth of slots was taken constant in the single and double damage scenarios and assumed variable in the triple ones. The vibration mode shapes together with their corresponding curvatures were obtained using the modal analysis. The contourlet transform coefficients of modal curvatures in two states of damaged and undamaged models were taken as the inputs for the proposed damage index. The results show that the proposed damage index has well identified the severity of triple damage scenarios in addition to detecting the location of different single and double damages at the middle and in the vicinity of corner and supports of the prestressed concrete slab models. Furthermore, the proposed damage index has the highest sensitivity rate to damage scenarios with geometric shapes of inclined, curved, transverse, and longitudinal slot, respectively.
Journal of Soft Computing in Civil Engineering
Pouyan Press
2588-2872
5
v.
2
no.
2021
39
67
https://www.jsoftcivil.com/article_133422_099a83bdd53119c94c56f5122bc58f17.pdf
dx.doi.org/10.22115/scce.2021.282138.1301
Assessing Compressive Strength of Concrete with Extreme Learning Machine
Sarat
Nayak
Department of Computer Science and Engineering, CMR College of Engineering & Technology, Hyderabad - 501401, India
author
Sanjib
Nayak
Department of Computer Application, VSS University of Technology, Odisha – 768018, India
author
Sanjay
Panda
Department of Computer Science and Engineering, National Institute of Technology, Warangal – 506004, India
author
text
article
2021
eng
Manual estimation of compressive strength of concrete (CSC) is time consuming and expensive. Soft computing techniques are found better to statistical methods applied to this problem. However, sophisticated prediction models are still lacking and need to be explored. Extreme learning machine (ELM) is a faster and better learning method for artificial neural networks (ANNs) with solitary hidden layer and has enhanced generalization capacity. This article presents an ELM-based forecast for efficient prediction of CSC. A publicly available dataset from UCI repository is used to develop and access the performance of the model. The prediction accuracy of ELM is compared with few machine learning methods such as back propagation neural network (BPNN), support vector machine (SVM), auto-regressive integrated moving average (ARIMA), and least squared estimation (LSE). A comparative study for the prediction of CSC at the curing ages of 28, 56, and 91 days has been carried out using all models. The experimental findings from ELM-based forecasting demonstrate its ability in predicting CSC in a robust manner. On an average, it achieves lowest MAPE of 0.048024, ARV of 0.052872, U of Theil’s statistics (UT) of 0.038772, NMSE of 0.058522, and standard deviation (SD) of 0.256267. Comparative analysis of simulation results and statistical significance test suggests the superiority of ELM-based CSC prediction.
Journal of Soft Computing in Civil Engineering
Pouyan Press
2588-2872
5
v.
2
no.
2021
68
85
https://www.jsoftcivil.com/article_133408_6dcee9cb0bb0e25be34fdf465dd906f1.pdf
dx.doi.org/10.22115/scce.2021.286525.1320
A Hybrid Generalized Reduced Gradient-Based Particle Swarm Optimizer for Constrained Engineering Optimization Problems
Hesam
Varaee
Assistant Professor, Department of Civil Engineering, Ale Taha Institute of Higher Education, Tehran, Iran
author
Naser
Safaeian Hamzehkolaei
Assistant Professor, Department of Civil Engineering, Bozorgmehr University of Qaenat, Qaen, Iran
author
Mahsa
Safari
MSc, Department of Civil Engineering, Ale Taha Institute of Higher Education, Tehran, Iran
author
text
article
2021
eng
A hybrid algorithm is presented that combines strong points of Particle Swarm Optimization (PSO) and Generalized Reduced Gradient (GRG) algorithm to keep a good compromise between exploration and exploitation. The hybrid PSO-GRG quickly approximates the optimum solution using PSO as a global search engine in the first phase of the search process. The solution accuracy is then improved during the second phase of the search process using the GRG algorithm to probe locally for a proper solution(s) in the vicinity of the current best position obtained by PSO. The k-nearest neighbors (k-NN)-based Purely Uniform Distributed (PUD) initial swarm is also applied to increase the convergence speed and reduce the number of function evaluations (NFEs). Hybridization between both algorithms allows the proposed algorithm to accelerate throughout the early stages of optimization using the high exploration power of PSO whereas, promising solutions will possess a high probability to be exploited in the second phase of optimization using the high exploitation ability of GRG. This prevents PUD-based hybrid PSO-GRG from becoming trapped in local optima while maintaining a balance between exploration and exploitation. The competence of the algorithm is compared with other state-of-the-art algorithms on benchmark optimization problems having a wide range of dimensions and varied complexities. Appraising offered algorithm performance revealed great competitive results on the Multiple Comparison Test (MCT) and Analysis of Variance (ANOVA) test. Results demonstrate the superiority of hybrid PSO-GRG compared to standard PSO in terms of fewer NFEs, fast convergence speed, and high escaping ability from local optima.
Journal of Soft Computing in Civil Engineering
Pouyan Press
2588-2872
5
v.
2
no.
2021
86
119
https://www.jsoftcivil.com/article_133423_56bc12f498d3ee4f3bf36a86e367074c.pdf
dx.doi.org/10.22115/scce.2021.282360.1304
Modelling of Daily Suspended Sediment Concentration Using FFBPNN and SVM Algorithms
Atul
Rahul
Assistant Professor, Faculty of Civil Engineering, MIT, Muzaffarpur, Bihar, Research Scholar IIT (BHU), Varanasi, India
author
Nikita
Shivhare
Assistant Professor, Faculty of Electronics and Communication Engineering, Oriental Group of Institutions, India
author
Shashi
Kumar
Assistant Professor, Faculty of Civil Engineering, SIT, Sitamarhi, Bihar, India
author
Shyam
Dwivedi
Professor, Faculty of Civil Engineering, IIT (BHU), Varanasi, India
author
Prabhat
Dikshit
Professor, Faculty of Civil Engineering, IIT (BHU), Varanasi, India
author
text
article
2021
eng
The river is an essential resource of fresh water on the earth, and its management is very challenging. Sedimentation and erosion is a very complex process of the river system. Suspended sediment concentration (SSC) plays a key role in this process. Therefore, water resources planning and management are essential for this. Generally, the sediment concentration estimated by direct measurement, but this process is costly and cannot apply in all rivers. It is essential to develop some technology that can predict the suspended sediment concentration. So, in this study, a feed forward back propagation neural network (FFBPNN) and support vector machine (SVM) were used to predict the suspended sediment concentration. One year of daily data was collected from the river Ganga at Varanasi cross-section. The performance of the model estimated for training and validation stages based on root mean square error (RSME), Coefficient of correlation (R) and Nash–Sutcliffe model efficiency (NSE). The performance of applied model indicated that FFBPNN (RSME = 176.2, R = 0.955 and NSE = 0.912) for validation is more precise for suspended sediment load prediction than SVM (RSME = 222.1, R = 0.930 and NSE = 0.864). This study shows that the soft computing technique is a robust tool for SSC prediction.
Journal of Soft Computing in Civil Engineering
Pouyan Press
2588-2872
5
v.
2
no.
2021
120
134
https://www.jsoftcivil.com/article_133932_afb5867d83a7bbc9eac02a19f58aea9f.pdf
dx.doi.org/10.22115/scce.2021.283137.1305
Forecasting the Shear Strength of Binary Blended Concrete Containing Hydrated Lime Using Artificial Intelligence
Chioma
Awodiji
Senior Lecturer, Department of Civil and Environmental Engineering, University of Port-Harcourt, Nigeria
author
David
Onwuka
Professor, Department of Civil Engineering, Federal University of Technology, Owerri, Nigeria
author
Samuel
Sule
Senior Lecturer, Department of Civil and Environmental Engineering, University of Port-Harcourt, Nigeria
author
text
article
2021
eng
In this exploratory study, the shear strength of blended cement concrete made using the hydrated lime (HL) as an admixture was studied. 120 shear strength values were experimentally obtained for several mix ratios at 7, 14, 21, and 28 days. This concrete was put together from water, portland cement (PC), HL, river sand (RS), and granite chippings (GC). 96 of the results were utilized to formulate a Levernberg-Marquardt backpropagation artificial neural network (ANN) for determining the shear strength of the concrete. The effectiveness of the forecast was tested using the unused 24 results. The model had 6 input variables namely; proportions of PC, HL, RS, GC, water, and curing age. While the output variable was the shear strength value. 1 hidden layer of 20 neurons was adopted. Uppermost 28 days shear strength value of 1.257N/mm2 was observed at 13.75% replacement of PC with HL for 0.58 water-cement ratio. The performance of the ANN proved that the model was acceptably executed. Root mean square errors (RMSE) obtained between network forecast and experimental values ranged from 0.0278 to 0.06536. These are close to 0. In addition, the factor of agreement (IA) determined were within the limits of 0.0475 and 0.1747. These are between the stipulated range of 0 to 1 for consistency between variables. The highest average percentage error recorded between model predictions and experimental values was 2.5066%. Lastly, the ANN created can be convincingly used to predict the shear strength of hydrated lime cement concrete and eliminate the need for try-out laboratory research.
Journal of Soft Computing in Civil Engineering
Pouyan Press
2588-2872
5
v.
2
no.
2021
135
151
https://www.jsoftcivil.com/article_134832_865abc0e564654ef8ba1dd5292b7f0bf.pdf
dx.doi.org/10.22115/scce.2021.275318.1285