Application of Gaussian Process Regression Model to Predict Discharge Coefficient of Gated Piano Key Weir

被引:56
作者
Akbari, Masood [1 ]
Salmasi, Farzin [1 ]
Arvanaghi, Hadi [1 ]
Karbasi, Masoud [2 ]
Farsadizadeh, Davood [1 ]
机构
[1] Univ Tabriz, Dept Water Engn, POB 5166616471, Tabriz, Iran
[2] Univ Zanjan, Dept Water Engn, Zanjan, Iran
关键词
Gated piano key (GPK) weir; Experimental model; Discharge coefficient (C-d); Gaussian process regression (GPR); Artificial intelligence; NEURAL-NETWORK; FLOW;
D O I
10.1007/s11269-019-02343-3
中图分类号
TU [建筑科学];
学科分类号
0813 ;
摘要
The Piano Key (PK) weir is a new type of long crested weirs. This study was involved the addition of a gate to PK weir inlet keys. It was conducted by the Department of Water Engineering, University of Tabriz, Iran to determine if the gate increased hydraulic performance. A Gated Piano Key (GPK) weir was constructed and tested for discharge ranges of between 10 and 130 l per second. To this end, 156 experimental tests were performed and the effective parameters on the GPK weir discharge coefficient (C-d), such as gate dimensions (b and d), gate insertion depth in the inlet key (H-gate), the ratio of the inlet key width to the outlet key width (W-i/W-o) and the head over the GPK weir crest (H) were investigated. In addition, application of soft computing to estimate of C-d was carried out using MLP, GPR, SVM, GRNN, multiple linear and non-linear regressions methods using MATLAB 2018 software. This study suggests the relation for C-d with non-dimension parameters. The results of this study showed that H, W-i/W-o, H-gate and b and d, had the greatest effect on the GPK weir discharge coefficient, respectively. The GPR method was introduced as a new effective method for predicting discharge coefficient of weirs with RMSE = 0.011, R-2 = 0.992 and MAPE = 1.167% and provided the best results when compared with other methods.
引用
收藏
页码:3929 / 3947
页数:19
相关论文
共 39 条
[32]   Gaussian Process Regression for Estimating Chlorophyll Concentration in Subsurface Waters From Remote Sensing Data [J].
Pasolli, Luca ;
Melgani, Farid ;
Blanzieri, Enrico .
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2010, 7 (03) :464-468
[33]   Neural network design for engineering applications [J].
Rafiq, MY ;
Bugmann, G ;
Easterbrook, DJ .
COMPUTERS & STRUCTURES, 2001, 79 (17) :1541-1552
[34]   Predicting discharge coefficient of compound broad-crested weir by using genetic programming (GP) and artificial neural network (ANN) techniques [J].
Salmasi, Farzin ;
Yildirim, Gurol ;
Masoodi, Azam ;
Parsamehr, Parastoo .
ARABIAN JOURNAL OF GEOSCIENCES, 2013, 6 (07) :2709-2717
[35]   Improved side weir discharge coefficient modeling by adaptive neuro-fuzzy methodology [J].
Shamshirband, Shahaboddin ;
Bonakdari, Hossein ;
Zaji, Amir Hossein ;
Petkovic, Dalibor ;
Motamedi, Shervin .
KSCE JOURNAL OF CIVIL ENGINEERING, 2016, 20 (07) :2999-3005
[36]  
Vapnik V., 2013, The nature of statistical learning theory
[37]   An online Bayesian filtering framework for Gaussian process regression: Application to global surface temperature analysis [J].
Wang, Yali ;
Chaib-draa, Brahim .
EXPERT SYSTEMS WITH APPLICATIONS, 2017, 67 :285-295
[38]   Support vector regression for modified oblique side weirs discharge coefficient prediction [J].
Zaji, Amir Hossein ;
Bonakdari, Hossein ;
Shamshirband, Shahaboddin .
FLOW MEASUREMENT AND INSTRUMENTATION, 2016, 51 :1-7
[39]   Landslide susceptibility mapping at Vaz Watershed (Iran) using an artificial neural network model: a comparison between multilayer perceptron (MLP) and radial basic function (RBF) algorithms [J].
Zare, Mohammad ;
Pourghasemi, Hamid Reza ;
Vafakhah, Mahdi ;
Pradhan, Biswajeet .
ARABIAN JOURNAL OF GEOSCIENCES, 2013, 6 (08) :2873-2888