文献类型：期刊文献

英文题名：Runoff Prediction Using a Novel Hybrid ANFIS Model Based on Variable Screening

作者：Liu, Zhennan Li, Qiongfang Zhou, Jingnan Jiao, Weiguo Wang, Xiaoyu

第一作者：刘振男

通信作者：Zhou, JN[1]

机构：[1]Sch Civil Engn, Guizhou Inst Technol, Guiyang 550003, Peoples R China;[2]Hohai Univ, Coll Hydrol & Water Resources, Nanjing 210098, Peoples R China;[3]Anhui Agr Univ, Sch Engn, Hefei 230036, Peoples R China

第一机构：贵州理工学院土木工程学院

通信机构：corresponding author), Sch Civil Engn, Guizhou Inst Technol, Guiyang 550003, Peoples R China.|贵州理工学院土木工程学院;贵州理工学院;

年份：2021

卷号：35

期号：9

起止页码：2921-2940

外文期刊名：WATER RESOURCES MANAGEMENT

收录：;EI(收录号：20212710591176);Scopus(收录号：2-s2.0-85109044586);WOS:【SCI-EXPANDED(收录号:WOS:000668027800001)】；

基金：This work is supported by the National Natural Science Foundation of China (No. 52069005), the Guizhou province science and technology fund (Guizhou Science Foundation-ZK[2021] General 295), and High-level Talents Start-up Fund Project of Guizhou Institute of Technology (XJGC20210425) and special thanks are given to the anonymous reviewers and editors for their constructive comments.

语种：英文

外文关键词：Monthly runoff forecasting; Adaptive neuro-fuzzy inference system; Fireworks algorithm; Uncertainty analysis

摘要：The accurate and reliable prediction of future runoff is important to guarantee for strengthening water resource optimization and management. The novel contribution of this article is the development of a hybrid model (FWA-ANFIS), which is based on the improvement of the adaptive neuro-fuzzy inference system (ANFIS) with the fireworks algorithm (FWA). The dominant driving factors of runoff are selected from several hydro-meteorological indices (precipitation, soil moisture content, and evaporation) as predictors by correlation coefficient (CC) analysis, mutual information (MI) analysis, correlation analysis and principal component analysis (CC-PCA), mutual information and kernel principal component analysis (MI-KPCA), MI-PCA, and CC-KPCA. The FWA-ANFIS model is applied to the Beiru River, China, with data from 1985-2016 (1985-2012 for model training and 2013-2016 for model prediction). The standard ANFIS, the GA-ANFIS, the PSO-ANFIS, the FWA-ELM, the GA-ELM, and the PSO-ELM are utilized as compared prediction models on the identical dataset. The results indicate that CC-PCA outperforms the other methods regarding the selection of predictors, and FWA-ANFIS has the best performance in terms of the root mean square error, correlation coefficient, and coefficient of determination, followed by the GA-ANFIS, PSO-ANFIS, ANFIS, FWA-ELM, GA-ELM, and PSO-ELM models. Furthermore, the degrees of uncertainty of the models increase in the following order: FWA-ANFIS, GA-ANFIS, PSO-ANFIS, ANFIS, PSO-ELM, GA-ELM, and FWA-ELM.