基于改进粒子群算法的神经网络建模

文章利用粒子群算法优化神经网络的参数，提出了基于粒子群算法的神经网络建模方法。为了提高基本粒子群算法的搜索性能，采用了基于外推技巧的引导型更新公式，并在粒子的搜索过程中，不断监测各个粒子的最优位置，多次没有变化并且距离优化目标太远时，粒子跳出当前位置继续搜索，从而避免陷入局部值。最后使用改进后的粒子群神经网络算法对函数进行拟合，仿真结果表明，新的算法有较好的收敛性。

The Modeling of Neural Network Based on Improved Particle Swarm Optimization

Particle swarm optimization is an optimization technique based on swarm intelligence, which is simple, easy to implement. In this paper, a modeling method combined PSO algorithm with neural network is proposed by using particle swarm optimization to optimize neural network parameters. To improve particle swarm optimization＇ s searching performance, a guided updating formula is adopted based on extrapolation techniques, meanwhile, the optimal location of each particle is constantly monitored in the searching process. if the location has not changed constantly and far to the optimal target, the particles jump out of the current position to continue the searching in order to avoid falling into local values. Finally, function approximation is simulated by using the improved PSO-NN algorithm. The results show that the new algorithm has better convergence.