Autor: |
Qiao, Dewen, Zhou, Pengzhan, Li, Mingyan, Guo, Songtao |
Předmět: |
|
Zdroj: |
Soft Computing - A Fusion of Foundations, Methodologies & Applications; Sep2024, Vol. 28 Issue 17/18, p9509-9520, 12p |
Abstrakt: |
Takagi–Sugeno fuzzy neural network (TSFNN) has been widely used in intelligent prediction. The prediction accuracy of TSFNN is impacted by its model parameter choices. However, the manual selection of parameters is hard to ensure the prediction accuracy of TSFNN. Therefore, we propose a Particle Swarm Optimization with Result Precision Enhancement strategy, PSO-RPE, to automatically generate highly optimized parameters to improve the prediction accuracy of TSFNN. In PSO-RPE, we present memory enhancement technology to direct the evolution of particles and accelerate the search process for the solution. Next, our PSO-RPE strategy employs a customized linear dynamic equation to balance the global and local search capabilities. Moreover, the elite strategy is applied to produce better feasible parameters with high probability. Moreover, the analysis of the convergence on 12 benchmarks is given to verify the superiority of PSO-RPE. Finally, our experiments in predicting water quality based on TSFNN demonstrate that the PSO-RPE strategy is effective in the generation of optimized parameters. [ABSTRACT FROM AUTHOR] |
Databáze: |
Complementary Index |
Externí odkaz: |
|