Application and Parameter Optimization of Dynamic Fuzzy Neural Network Based on Improved Chaotic Particle Swarm Optimization

The performance and the study stability of dynamic fuzzy neural network (DFNN) depend on its preset parameters selection. To the multi-parameter optimization problem of DFNN, improved chaotic particle swarm optimization (ICPSO) is proposed to find the best combination of the preset parameters of DFNN. The experimental results further demonstrate that the ICPSO provides an effective way to search the best parameters combination of DFNN, and has the better precision, the quicker convergence speed and the fewer iteration steps. DFNN model based on improved chaotic particle swarm optimization is built for coal and gas outbursts, which has good modeling and higher prediction accuracy.