Paper Info
Title
Time Series Prediction Based On Echo State Network Tuned By Divided Adaptive Multi-Objective Differential Evolution Algorithm
Abstract
For echo state networks, it is difficult to select suitable reservoir parameters for different applications. In this paper, we put forward a divided adaptive multi-objective differential evolution (DAMODE) algorithm to optimize the reservoir parameters of echo state network. To improve the performance of multi-objective differential evolution algorithm, the entire population is divided into several subpopulations, and each subpopulation is divided into two subsets to compromise convergence and diversity, which are updated according to certain rules. Besides, the scale factor and crossover rate of differential evolutionary algorithm are adaptively adjusted. Experiments were conducted on the Lorenz time series, hourly temperature time series and PM2.5 time series in Beijing. Experiment results show that the proposed model can improve prediction accuracy and has good generalization ability and practicability.
Year
DOI
Venue
2021
10.1007/s00500-020-05457-8
SOFT COMPUTING
Keywords
DocType
Volume
Time series, Prediction, Multi-objective differential evolution, Echo state network
Journal
25
Issue
ISSN
Citations 
6
1432-7643
0
PageRank 
References 
Authors
0.34
0
3
Name
Order
Citations
PageRank
Weijie Ren1385.41
Yiwen Wang200.34
Min Han376168.01