基于粒子群算法的BP神经网络优化技术

摘要

针对传统BP神经网络存在学习效率低、收敛速度慢和容易陷入局部极小值的问题，提出一种基于改进的PSO来优化BP神经网络的方法。通过在PSO算法中引入随机变化的加速常数来获得最优权值，对BP神经网络进行优化和训练，将优化的BP神经网络用于遗传高血压患病年龄的预测中。实验结果表明，该方法较好地解决了传统BP神经网络易陷入局部极小值的问题，提高了算法的收敛速度和稳定性。%Aiming at the problems that traditional BP neural network learning is inefficient,has slow convergence and is easy to fall into local minimum value,a method which was based on improved PSO optimizing the BP neural network was proposed.By introducing the random variation acceleration constant in the PSO algorithm,the optimal weights were obtained.And then the BP neural network was optimized and trained which was used to forecast genetic hypertension in age.The experimental results show that the optimized BP neural network solves the problem of easily trapping into local minimum.At the same time,the method improves the speed and stability of the convergence of the algorithm.