The Use of Neural Networks in Determining the Discharge Coefficient of Non-Linear Arc Weir

tohid omidpour alavian¹ mehdimajediasl² ahadmehrabi³ mahdisoltani⁴

1) Civil Engineering, University of Maragheh, Maragheh, Iran
2) Department of Civil Engineering, Faculty of Engineering, University of Maragheh, Maragheh, Iran.
3) Civil Engineering, University of Maragheh, Maragheh, Iran
4) Civil Engineering, University of Maragheh, Maragheh, Iran

IOI Code : XYCF-KABFE
Publication : 2nd.International Congress on Management, Economy, Humanities and Business Development(icmba.ir)
Abstract :
    Weirs are one of the common structures in many dams and water transfer channels. Weirs are always considered as one of the important components of water transmission and distribution networks, which despite their simple structure and function, play an important role in these networks and also for the passage of excess water from upstream to downstream. A structure called a weir is used in dams. Smart algorithms have found a very valuable position among researchers due to their great ability to discover complex and secret relationships between effective independent parameters and dependent parameters, as well as saving money and time. In this research, the performance of SVM and QNET algorithms in predicting the flow rate coefficient of non-linear arc weir has been investigated with the help of 243 laboratory data series. The geometric and hydraulic parameters used in this research include water load (HT), weir height (P), total water load ratio (H_T/p), arc cycle angle (Ɵ), cycle wall angle (α) and coefficient It is discharge (Cd). The results of artificial intelligence showed that the combination of parameters (Cd, H_T/p, α, Ɵ) in the SVM algorithm in the training and test phase are respectively equal to (R2=0.9907), (RMSE=0.0183), (DC=0.9871) and (R2=0.9914), (RMSE=0.0206), (DC=0.9848) and in the QNET algorithm in the training and test phase respectively equal to (R2=0.9876), (RMSE=0.0215), (DC=0.9878) and (R2=0.9637), (RMSE=0.0736), (DC=0.8967) compared to other combinations has led to the most optimal output, which indicates the very favorable accuracy of SVM and QNET algorithms in predicting the flow rate coefficient of non-linear arc weir. The results of the sensitivity analysis showed that the effective parameter in determining the non-linear arc weir discharge coefficient in SVM and QNET is the total water load ratio parameter (H_T/p).

Keywords : non linear weirs discharge coefficient sensitivity analysis neural networks