机构地区: 武汉光电国家实验室
出 处: 《机械工程师》 2011年第9期74-77,共4页
摘 要: 分析了粒子群算法(PSO)和差分进化算法(DE)的特点,提出了一种PSO-DV算法用于优化BP神经网络的权值和阈值,并利用PSO-DV算法优化的BP神经网络进行了齿轮箱的故障诊断。试验结果表明,PSO-DV算法可以避免神经网络陷入局部极小,改善了收敛性能,同时保证了齿轮箱故障诊断的正判率。 A PSO-DV algorithm is proposed based on analysis of characteristic of particle swarm optimization and differential evolution algorithms to optimize parameters of BP neural network. Then, the PSO-DV algorithm trained BP neural network is applied to a gear-box fault diagnosis experiment. The experimental result indicates that the BP neural network training method escape from local minimum value using PSO-DV algorithm. Meanwhile, based on the PSO algorithm is an effective training algorithm, and it is also an available approach to solve fault diagnosis problems.
领 域: [兵器科学与技术]