| Cite this article: | 李守涛,王军涛,于明,等.基于GA-RBF神经网络的位山闸引水能力预测研究[J].灌溉排水学报,0,():-. |
| LI Shoutao,WANG Juntao,YU Ming,et al.基于GA-RBF神经网络的位山闸引水能力预测研究[J].灌溉排水学报,0,():-. |
|
| |
|
|
| This article has been：Browse 1399Times Download 0Times |
|
|
|
| DOI： |
|
| Study on Water Diversion Capability of Weishan SluiceBased on GA-RBF Neural Network |
|
LI Shoutao1, WANG Juntao2, YU Ming3, YAO Jingwei2, ZHAO Guoping4, FAN Yumiao2
|
|
1..Hohai University College of Water Conservancy and Hydropower Engineering;2.Yellow River Institute Of Hydraulic Research;3.Weishan Irrigation District Management Office of Liaocheng City;4.Yellow River Affairs Bureau of Liaocheng City
|
| Abstract: |
| Due to the riverbed undercut and river regime changes,etc, the Weishan Sluice of the lower Yellow River has a large decrease in water diversion capacity at the same flow level. In addition, there are hidden safety hazards,so it is urgently needed to rebuild. Both the design of the reconstruction project and the operation and management of the irrigation area require prediction of the culvert's water diversion capacity. Traditional hydraulic empirical formulas predicting has problems such as unavailability of some data, complex data measurement, and the need to manually modify parameters,which make the operation inconvenient and inflexible.【Objective】In order to make a scientific prediction of the water diversion capacity after the Weishan Sluice reconstruction.【Method】Based on the advantages of strong nonlinear fitting ability of Radiical Basis Function (RBF) neural network and strong optimization ability of genetic algorithm,this paper establishs a Weishan Sluice water diversion capacity prediction model based on genetic algorithm (GA) to optimize the parameters of the hidden layer of the RBF neural network.,the input variables of the model are the number of sluice openings, the water depth before and after the sluice, and the seasonal factor (affected by the flood season, the seasonal factor is 1 during the flood season, and the seasonal factor is 2 during the non-flood season), and the output variables are the measured sluice flow. It uses the sample set composed of current water regime data to train and test the model. The average error of the trained model is 1.64%, which proves that the prediction effect is better. 【Result】The trained model was used to predict the water diversion capacity of the reconstructed mountain gate, and the results show that in the flood season, the water diversion capacity can meet the design requirements and the water diversion capacity increases with the reduction of the head after the sluice in the non-flood season, considering the diversion scheme of the sand conveyance channel after the sluice, it can basically meet the design requirements. 【Conclusion】The GA-RBF model has strong adaptability and high prediction accuracy in predicting the water diversion capacity of Weishan Sluice, and has certain value for popularization and application. |
| Key words: Weishan Sluice;GA-RBF;diversion capacity;prediction |
|
|
|
|
|
|
