Time series prediction model of echo state network based on regular octahedral topological structure
ZHANG Yuanyuan
LUN Shuxian
Abstract：Echo state networks are widely used in chaotic time series prediction due to their superior performance.However,in practice,the complex topology and random reservoir of echo state networks are difficult to implement.An echo state network model based on sulfur hexafluoride octahedral topology is proposed to address this issue.The model has the following characteristics:the fixed connection between neurons in the reservoir improves the internal stability of the reservoir while reducing the randomness and blindness between neurons,and increases the efficiency of information transmission.Simulation tests were conducted using sunspot time series for prediction,and validated with Indian stock data.The experimental results showed that the improved model has higher prediction accuracy.
Keywords：echo state networktime series predictionregular octahedral topological structure
Publication Date：2025-06-15
Online Publishing Date：2025-09-09(First online date of this platform, not the publication date of the document)
Pages：8( 153-160 )