摘要
论述了一种改进的鱼群算法,利用其全局寻优能力优化BP神经网络的权值和阈值,形成一套基于改进鱼群算法优化神经网络的故障诊断方法(ADAFSA-BP)。通过试验采集和处理轴承故障信息,应用GA-BP,SFLA-BP和ADAFSA-BP对试验数据进行处理和对比分析,结果表明:ADAFSA-BP不仅加快了神经网络的收敛速度,而且在诊断精度上有了较大提高。
An improved fish swarm algorithm is discussed. The global optimization ability is used to optimize weights and thresholds of BP neural network, and a fault diagnosis method(ADAFSA-BP) is developed. The fault information for bearings is collected and processed by experiment, and the experimental data are processed and comparatively analyzed by GA-BP, SFLA-BP and ADAFSA-BP. The results show that the convergence speed of neural network is accelerated and the diagnostic accuracy is improved greatly by ADAFSA-BP.
引文
[1] 王国彪,何正嘉,陈雪峰,等.机械故障诊断基础研究“何去何从”[J].机械工程学报,2013,49(1):63-72.
[2] 王立臣,梁浩.滚动轴承故障诊断技术现状及发展趋势[J].电子测试,2013(22):141-143.
[3] LOUSSIFI H ,NOURI K,BRAIEK N B.A new efficient hybrid intelligent method for nonlinear dynamical systems identification:the wavelet kernel fuzzy neural network[J].Communications in Nonlinear Science and Numerical Simulation,2015:S1007570415002877.
[4] 刘浩然,赵翠香,李轩,等.一种基于改进遗传算法的神经网络优化算法研究[J].仪器仪表学报,2016,37(7):1573-1580.
[5] 李晓磊.一种新型的智能优化方法-人工鱼群算法[D].杭州:浙江大学,2003.
[6] 龚波,曾飞艳.一种改进人工鱼群算法对BP神经网络的优化研究[J].湖南科技大学学报(自然科学版),2016,31(1):86-90.
[7] 高宇航.一种改善BP神经网络性能的方法研究[J].微型机与应用,2017,36(6):53-57,61.
[8] YAO Z T,PAN H X.Engine fault diagnosis based on improved BP neural network with conjugate gradient[J].Applied Mechanics & Materials,2014 (536/537):296-299.
[9] 朱云博,周厚强,张燕军,等.基于多方法结合应用的齿轮箱故障特征提取研究[J].山东理工大学学报(自然科学版),2013,27(5):76-78.
[10] 魏秀业.基于粒子群优化的齿轮箱智能故障诊断研究[D].太原:中北大学,2009.