ELM与BP神经网络模型在径流预报中的比较研究  

Comparative Study of ELM and BP Neural Network Models for Runoff Prediction

作　　者：王文川 李文锦 徐冬梅 李庆敏 

机构地区：[1]华北水利水电大学水利学院,河南郑州

出　　处：《水资源研究》2018年第6期551-556,共6页Journal of Water Resources Research

基　　金：国家自然科学基金项目(51509088);河南省高校科技创新团队(14IRTSTHN028);河南省水环境模拟与治理重点实验室(2017016)

摘　　要：为了更加精确的进行径流预测,该研究针对BP神经网络训练速度慢和容易陷入局部极小值的缺点,建立了ELM神经网络模型。以兰西水文站1959~2014年径流数据为例,采用ELM神经网络对径流深进行预测,相对误差、均方误差和确定性系数作为模型合理性的验证指标,并与BP神经网络预测结果进行对比及分析。ELM模型的预测结果,其相对误差、均方误差、确定性系数均优于BP神经网络模型,这表明ELM神经网络模型对BP神经网络模型已存在缺点进行了有效规避且预测精度有了进一步的提升。因此,该研究提供的ELM模型在一定程度上能够更好的改善预测效果,证明了ELM模型在径流预报中的应用价值。In order to make the runoff prediction more accurate, this study established the ELM neural network model for the shortcomings of BP neural network training slow and easy to fall into local minimum. Taking the runoff data of Lanxi Hydrological station from 1959 to 2014 as an example, the ELM neural network predicts the runoff depth. The relative error, mean square error and decision coefficient are used as the verification indicators of the rationality of the model, and compared with the BP neural network prediction results. The prediction results show that the ELM model is better than BP neural network model in terms of relative error, mean square error and decision coefficient. This indicates that the ELM neural network model has effectively avoided the shortcomings of the BP neural network model and the prediction accuracy has been further improved. Therefore, the ELM model can improve the prediction effect to a certain extent which has application value in annual runoff prediction.

关 键 词：神经网络 ELM模型 BP模型 径流预测 

分 类 号：TP1[自动化与计算机技术—控制理论与控制工程]