计算机科学
支持向量机
稳健性(进化)
数据挖掘
水质
人工神经网络
人工智能
算法
机器学习
生态学
生物化学
生物
基因
化学
作者
Chenguang Song,Lei Yao,Chengya Hua,Qihang Ni
标识
DOI:10.1016/j.jhydrol.2021.126879
摘要
Predicting water quality accurately is of utmost importance in water resource management. This study focuses on the prediction of water quality parameters such as the dissolved oxygen (DO) in the watershed system. The traditional recurrent neural networks (RNNs) suffer from gradient disappearance or explosion and inability to solve the problem of long-time dependence, and their practical applications are limited. To overcome this drawback, an improved long short-term memory (LSTM) model was proposed to improve the model’s performance. In addition, to overcome nonstationarity, randomness, and nonlinearity of the water quality parameters data, a hybrid model that recruits synchrosqueezed wavelet transform (SWT) to denoise the original data was adopted. A novel metaheuristic optimization algorithm, the improved sparrow search algorithm (ISSA) combining Cauchy mutation and opposition-based learning (OBL) was also implemented to compute the optimal parameter values for the LSTM model. The proposed hybrid model was evaluated using an original weekly water quality parameters series, from 1/2010–12/2016, measured at the Yongding River and Gangnan gauging stations in the Haihe River Basin, China. The standalone LSTM, ISSA-LSTM, SWT-LSTM, support vector regression (SVR), and back propagation neural network (BPNN) were adopted as comparative forecast models using the same dataset. The results demonstrate that the addressed model, combining the strong noise-resistant robustness of the SWT and the nonlinear mapping of the LSTM, has the best prediction performance among the peer models at two gauging stations. The proposed hybrid model can be used as an alternative framework for water quality prediction, which can provide decision-making basis for comprehensive water quality management and pollutant control in the basin.
科研通智能强力驱动
Strongly Powered by AbleSci AI