Mohammadmehdi Razmi; Mojtaba Saneie; Shamsa Basirat
Abstract
In this paper, the ANFIS network was optimized using three algorithms comprising the Particle Swarm Optimization (PSO), Firefly Algorithm (FFA), and Genetic algorithm (GA) for the first time. To ameliorate the ability of the numerical models,the Monte Carlo simulations were utilized. Moreover, in order ...
Read More
In this paper, the ANFIS network was optimized using three algorithms comprising the Particle Swarm Optimization (PSO), Firefly Algorithm (FFA), and Genetic algorithm (GA) for the first time. To ameliorate the ability of the numerical models,the Monte Carlo simulations were utilized. Moreover, in order to assess the simulation outcomes, the k-fold cross validation technique was implemented. Initially, using all inputs, five different parameters were used for producing theANFIS, ANFIS-GA, ANFIS-PSO, and ANFIS-FFA methods. After that, a computational fluid dynamics (CFD) model simulated the discharge coefficient (DC) and the outcome of all simulations were compared. The analysis of the results demonstrated that the ANFIS-FFA model approximates the DC with higher precision. For instance, the amount of the coefficient of determination and the scatter index were surmised as 0.961 and 0.039. Also, the side weir height ratio tothe upstream depth (P/y1) was detected as the most influential parameter. About 85% of the DC simulated by the ANFIS-FFA model had an inaccuracy of less than 5%. The performed uncertainty analysis proved that the best model possesses an underestimated efficiency. For this model, the influence of the inputs were analyzed in a ±10% range. Finally, a computational code was presented for the simulation of DC by hydraulic and environmental engineers.
Amir Hosein Azimi; Saeid Shabanlou; Behrouz Yaghoubi
Abstract
In this paper, the hydraulic jump length on a slope rough floor is simulated through the extreme learning machine (ELM). Then, the parameters affecting the hydraulic jump on the slope rough bed are detected. After that, five different ELM model are developed so as to determine the influenced factor. ...
Read More
In this paper, the hydraulic jump length on a slope rough floor is simulated through the extreme learning machine (ELM). Then, the parameters affecting the hydraulic jump on the slope rough bed are detected. After that, five different ELM model are developed so as to determine the influenced factor. Next, the results obtained from different ELM models are analyzed. The comparison of the results with the experimental data proves the acceptable accuracy of the mentioned numerical models. Regarding the results from the numerical method, the superior ELM model estimates the hydraulic jump length in terms of the flow Froude number, the ratio of bed roughness, the ratio of sequent depths and bed slope. The values of the root mean square error (RMSE), mean absolute percent error (MAPE), scatter index (SI) and correlation coefficient (R) for the superior model are respectively obtained 0.657, 3.507, 0.052 and 0.985. Based on the simulation, the flow Froude number at upstream is introduced as the most effective parameter in predicting the jump length on the sloping rough floor.
Mahdi Majedi-Asl; Rasoul Daneshfaraz; Mehdi Fuladipanah; John Abraham; Mohammad Bagherzadeh
Abstract
In this paper, two groups of datasets including Froehlich (1988) and USGS were implemented to simulate scour depth for bridge piers of three shapes (circular, sharp-nose and rectangular) using support vector machine (SVM) algorithm. The results of the SVM were compared with gene expression programming ...
Read More
In this paper, two groups of datasets including Froehlich (1988) and USGS were implemented to simulate scour depth for bridge piers of three shapes (circular, sharp-nose and rectangular) using support vector machine (SVM) algorithm. The results of the SVM were compared with gene expression programming (GEP) and the non-linear regression model. Independent parameters extracted using dimensional analysis were Froud number (Fr), the ratio of pier length to pier width (L/b), the ratio of sediment particle diameters (d50/d84), the ratio of sediment mean size to pier width (d50/b) and attack angle of water flow (α). Different combinations of independent variables were used to achieve optimum performance of the simulator. The results showed that among three simulators, the SVM algorithm had the best performance to predict local scour depth. The sensitivity analysis revealed that among independent parameters, the descending order of effectivity was Fr, sediment size, L/b, and α.
Mohammad Ali Izadbakhsh; Reza Hajiabadi
Abstract
In the article, through the adaptive neuro-fuzzy inference system (ANFIS), a sensitivity analysis is conducted on the variables affecting the discharge capacity of the weir. To this end, the variables affecting the discharge capacity of labyrinth weirs are initially identified. Then, using these input ...
Read More
In the article, through the adaptive neuro-fuzzy inference system (ANFIS), a sensitivity analysis is conducted on the variables affecting the discharge capacity of the weir. To this end, the variables affecting the discharge capacity of labyrinth weirs are initially identified. Then, using these input parameters, seven ANFIS models are developed for conducting the sensitivity analysis. After that, the most optimal membership function number for the ANFIS model is chosen. In other words, by conducting the trial and error process, the best number of the membership functions in terms of time and modeling accuracy are selected. Then, the sensitivity analysis is performed for the ANFIS models and the superior ANFIS model is chosen finally. The accuracy of the superior model in both the validation and testing artificial intelligence (AI) methods is in an acceptable range. For example, the scatter index (SI), correlation coefficient (R) and the Nash-Sutcliff efficiency coefficient (NSC) for the model in the testing mode are obtained 0.049, 0.964 and 0.924, respectively. It should be noticed that the outcomes of the sensitivity analysis show that the ratio of the weir head to the weir crest and the Froude number are introduced as the most effective input parameters. Eventually, a computer code is proposed to estimate the discharge capacity of labyrinth weirs by this model.
Ehsan Yarmohammadi; Fariborz Yosefvand; Ahmad Rajabi; Saeid Shabanlou
Abstract
In this paper, for the first time, the discharge coefficient of triangular plan form weirs is simulated by the extreme learning machine (ELM). ELM is one of the powerful and rapid artificial intelligence methods in modeling complex and non-linear phenomena. Compared to other learning algorithms such ...
Read More
In this paper, for the first time, the discharge coefficient of triangular plan form weirs is simulated by the extreme learning machine (ELM). ELM is one of the powerful and rapid artificial intelligence methods in modeling complex and non-linear phenomena. Compared to other learning algorithms such as back propagation, this model acts rapidly in the learning process and provides a desirable performance in processing generalized functions. In this study, the Monte Carlo simulation is used for examining capabilities of numerical models. Also, the k-fold cross validation method with k=5 is utilized for evaluating abilities of the ELM models. Then, six ELM models are introduced by means of the parameters affecting the discharge coefficient of triangular plan form weirs. After that, the superior model is identified by analyzing the results of the mentioned models. The superior model predicts discharge coefficient values with reasonable accuracy. This model simulates the discharge coefficient as a function of the flow Froude number, vertex angle of the triangular plan form weir, the ratio of weir length to its height, the ratio of flow head to weir height and the ratio of channel width to weir length. For the best model, the Mean Absolute Error, Root Mean Square Error and determination coefficient are computed 1.173, 0.012 and 0.967, respectively. Furthermore, examination of the influence of the input parameters indicates that the flow Froude number is the most influenced factor in modeling the discharge coefficient. Also, the error distribution showed that roughly 86 % of the superior model results had an error less than 2 %. Furthermore, a practical equation was provided to compute the discharge coefficient.
Mohammadali Izadbakhsh; Reza Hajiabadi
Abstract
In this paper, the discharge coefficient of weirs is simulated by the extreme learning machine (ELM). To this end, seven different ELM models are introduced by the input parameters. Also, the most optimal number of the neurons in the hidden layer is computed 7. Furthermore, different activation functions ...
Read More
In this paper, the discharge coefficient of weirs is simulated by the extreme learning machine (ELM). To this end, seven different ELM models are introduced by the input parameters. Also, the most optimal number of the neurons in the hidden layer is computed 7. Furthermore, different activation functions of the ELM model are assessed and the sigmoid activation function is taken into account as the most optimal one. Besides, the seven defined ELM models are analyzed and the superior model is introduced. This model approximates the discharge capacity with better performance in comparison with the other ELM models. It should also be noted that the superior ELM model is in terms of the dimensionless factors including Fr, HT/P, Lc/W, A/w, w/P. For the superior ELM model, the R2, VAF and NSC are respectively estimated 0.897, 89.626 and 0.892. Furthermore, the MAE and RMSE statistical indices for the ELM model are respectively estimated 0.024 and 0.031. Also, the most effective input parameters for modeling the discharge capacity of labyrinth weirs using the ELM are detected through the conduction of a sensitivity analysis, meaning that the HT/P is identified as the most influenced input parameter. Lastly, an applicable equation for computing the discharge capacity of labyrinth weirs is suggested which can be used by hydraulic and environmental engineers.
Azam Akhbari; Amir Hossein Zaji; Hamed Azimi; Mohsen Vafaeifard
Abstract
Weirs are installed on open channels to adjust and measure the flow. Also, discharge coefficient is considered as the most important hydraulic parameter of a weir. In this study, using the Radial Base Neural Networks (RBNN) and M5' methods, the discharge coefficient of triangular plan form weirs is modeled. ...
Read More
Weirs are installed on open channels to adjust and measure the flow. Also, discharge coefficient is considered as the most important hydraulic parameter of a weir. In this study, using the Radial Base Neural Networks (RBNN) and M5' methods, the discharge coefficient of triangular plan form weirs is modeled. At first, the effective parameters in the prediction of the discharge coefficient are identified. Then, by combining the input parameters, for each of the RBNN and M5' methods, six different models are introduced. By analyzing the modeling results for all models, it was shown that the M5' model is capable of modeling the discharge coefficient more accurately. Also, based on the modeling results, a model that considered the impact of all input parameters was introduced as a superior model. The mean absolute percentage error (MAPE) and correlation coefficients (R2) values for the preferred model in the test mode were calculated 2.774 and 0.831, respectively. Also, for each of the M5' models, some relationships were proposed to estimate the triangular plan form weirs. The evaluation of these relationships showed that the parameters of the ratio of head over the weir to channel width (h/B) and Froude number (Fr) were the most effective parameters in the prediction of the discharge coefficient.
