﻿ 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

Journal of Water Resources Research
Vol. 07  No. 06 ( 2018 ), Article ID: 27900 , 6 pages
10.12677/JWRR.2018.76062

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

Wenchuan Wang, Wenjin Li, Dongmei Xu, Qingmin Li

School of Water Conservancy, North China University of Water Resources and Electric Power, Zhengzhou Henan

Received: Nov. 15th, 2018; accepted: Nov. 30th, 2018; published: Dec. 7th, 2018

ABSTRACT

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.

Keywords:Neural Network, ELM Model, BP Model, Runoff Prediction

ELM与BP神经网络模型在径流预报中的比较研究

1. 引言

2. 模型介绍

2.1. ELM算法介绍

ELM是黄广斌教授提出的一种单隐含层前馈神经网络 [11] 。ELM的出现有效的解决了前馈神经网络学习速度慢的难题。该算法只需在训练之前随机生成输入层与隐含层之间的连接权值和隐含层神经元阈值，且训练过程中无需改动 [12] [13] 。

$\underset{j=1}{\overset{M}{\sum }}{\beta }_{i}g\left({w}_{i}\cdot {x}_{i}+{b}_{i}\right)={o}_{j}\text{\hspace{0.17em}}\text{\hspace{0.17em}}\text{\hspace{0.17em}}\left(j=1,2,\cdots ,N\right)$ (1)

$T=H\beta$ (2)

HT为输出权值矩阵的伪逆。

2.2. BP神经网络算法介绍

BP神经网络采用误差反向传播的思想，它实质上包含两个阶段，正向传播和反向传播，正向传播输出层得不到期望输出时，则进行反向传播进而对网络的各层连接权进行修正 [14] [15] 。建模过程如下：① 初始化网络权值和阈值；② 构建网络；③ 计算输出层误差；④ 反向修正权值；⑤ 重复③、④直到达到终止循环要求；⑥ 基于上述网络利用仿真函数进行数据预测 [16] 。再将预测数据反归一化就得到了最终的网络预测输出。

3. 实例应用

3.1. 研究区概况

3.2. 数据来源

3.3. 基于ELM模型与BP神经网络模型的预测

Table 1. Input and output parameters of the model

3.4. ELM模型与BP神经网络模型预测结果对比

Table 2. Comparison of ELM model and BP neural network model prediction results

3.5. 结果分析

Figure 1. Results comparison chart

4. 结论

Comparative Study of ELM and BP Neural Network Models for Runoff Prediction[J]. 水资源研究, 2018, 07(06): 551-556. https://doi.org/10.12677/JWRR.2018.76062

1. 1. 李红波, 夏潮军, 王淑英. 中长期径流预报研究进展及发展趋势[J]. 人民黄河, 2012, 34(8): 36-38. LI Hongbo, XIA Chaojun and WANG Shuying. Research progress and development trend of medium and long term runoff forecasting. People’s Yellow River, 2012, 34(8): 36-38. (in Chinese)

2. 2. 崔东文. 改进Elman神经网络在径流预测中的应用[J]. 水利水运工程学报, 2013(2): 71-77. CUI Dongwen. Application of improved Elman neural network in runoff prediction. Journal of Water Resources and Hydropower Engineering, 2013(2): 71-77. (in Chinese)

3. 3. 田雨波, 陈风, 张贞凯. 混合神经网络技术[M]. 科学出版社, 2015. TIAN Yubo, CHEN Feng and ZHANG Yukai. Hybrid neural network technology. Science Press, 2015. (in Chinese)

4. 4. 张正, 秦雨, 李聪, 等. 基于BP神经网络的尼尔基水库水质评价[J]. 水资源研究, 2017(3): 247-253. ZHANG Zheng, QIN Yu, LI Cong, et al. Water quality evaluation of Nierji Reservoir based on BP neural network. Water Resources Research, 2017(3): 247-253. (in Chinese)

5. 5. HUANG, G. B., ZHU, Q. Y. and SIEW, C. K. Extreme learning machine: A new learning scheme of feedforward neural networks. IEEE International Joint Conference on Neural Networks, 2005(2): 985-990.

6. 6. 赵坤, 覃锡忠, 贾振红, 王哲辉, 牛红梅. 基于CS算法改进ELM的时间序列预测[J]. 计算机工程与设计, 2018, 39(8): 2649-2653. ZHAO Kun, YAN Xizhong, JIA Zhenhong, WANG Zhehui and NIU Hongmei. Time series prediction based on improved ELM based on CS algorithm. Computer Engineering and Design, 2018, 39(8): 2649-2653. (in Chinese)

7. 7. 黄永辉, 李胜林, 樊祥伟, 等. ELM神经网络爆堆形态预测模型的研究及应用[J]. 煤炭学报, 2012, 37(s1): 65-69. HUANG Yonghui, LI Shenglin, FAN Xiangwei, et al. Research and application of ELM neural network popping morphological prediction model. Journal of China Coal Society, 2012, 37(s1): 65-69. (in Chinese)

8. 8. 姜媛媛, 刘柱, 罗慧, 等. 锂电池剩余寿命的ELM间接预测方法[J]. 电子测量与仪器学报, 2016, 30(2): 179-185. JIANG Yuanyuan, LIU Zhu, LUO Hui, et al. ELM indirect prediction method for residual life of lithium batteries. Journal of Electronic Measurement and Instrument, 2016, 30(2): 179-185. (in Chinese)

9. 9. 陈恒志, 杨建平, 卢新春, 等. 基于极限学习机(ELM)的连铸坯质量预测[J]. 工程科学学报, 2018(7). CHEN Hengzhi, YANG Jianping, LU Xinchun, et al. Quality prediction of continuous casting billet based on Extreme Learning Machine (ELM). Journal of Engineering Science, 2018(7). (in Chinese)

10. 10. 石炜, 师伟, 秦波. 基于ELM的高炉喷煤量预报研究[J]. 内蒙古科技与经济, 2017(13): 70-71. SHI Wei, SHI Wei and QIN Bo. Research on prediction of blast furnace coal injection based on ELM. Inner Mongolia Science and Technology and Economy, 2017(13): 70-71. (in Chinese)

11. 11. HUANG, G. B., ZHU, Q. Y. and SIEW, C. K. Extreme learning machine: Theory and applications. Neurocomputing, 2006, 70(1): 489-501.

12. 12. 叶智峰, 曹青. 极限学习机在初始地应力场反演中的应用[J]. 水电能源科学, 2016(6): 158-160. YE Zhifeng, CAO Qing. Application of extreme learning machine in initial geostress field inversion. Hydroelectric Energy Science, 2016(6): 158-160. (in Chinese)

13. 13. 秦琼, 刘树洁, 赖旭, 刘霄. GA优化ELM神经网络的风电场测风数据插补[J]. 太阳能学报, 2018, 39(8): 2125-2132. QIN Qiong, LIU Shujie, LAI Xu and LIU Wei. The wind data interpolation of wind farms based on GA optimized ELM neural network. Acta Energiae Sinica, 2018, 39(8): 2125-2132. (in Chinese)

14. 14. 车骞, 王根绪, 畅俊杰, 等. 基于人工神经网络的黄河源区枯季径流预报[J]. 人民黄河, 2005, 27(3): 23-24. CHE Wei, WANG Genxu, CHANG Junjie, et al. Forecast of dry season runoff in the source region of the Yellow River based on artificial neural network. People’s Yellow River, 2005, 27(3): 23-24. (in Chinese)

15. 15. 陈昕, 姚晓磊, 鱼京善. 基于BP人工神经网络模型的沙柳河流域径流模拟后处理研究[J]. 水资源研究, 2016(4): 391-401. CHEN Wei, YAO Xiaolei and YU Jingshan. Research on post-processing of runoff simulation in Shaliu River Basin based on BP artificial neural network model. Water Resources Research, 2016(4): 391-401. (in Chinese)

16. 16. 王博, 顿新春, 李智勇. 基于BP神经网络的水利工程投标决策模型及应用[J]. 水电能源科学, 2013(3): 131-134. WANG Bo, DENG Xinchun and LI Zhiyong. Bidding decision model and application of hydraulic engineering based on BP neural network. Hydroelectric Energy Science, 2013(3): 131-134. (in Chinese)

17. NOTES

作者简介：王文川(1976-)，男，河南鹿邑人，博士，教授，博导,主要从事水文水资源系统分析、遥感信息处理等方面的研究。