基于ANCEEMD样本熵特征提取的行星齿轮箱智能故障诊断

以行星齿轮箱为研究对象,针对经验模态分解(EMD)存在模态混叠、易出现端点效应等缺陷,导致难以很好地解决行星齿轮箱振动信号耦合及非线性特征的提取问题。本文以自适应噪声完备总体经验模态分解(ANCEEMD)为信号处理方法,引入样本熵进行特征提取,应用群智能融合算法优化的神经网络模型对行星齿轮箱故障进行识别和诊断。对于混合蛙跳算法(SFLA)与粒子群优化算法(PSO),实施“两层优化和内外循环”的融合机制,提出SFLA-PSO融合算法。开展了行星齿轮模拟故障实验,采集了行星齿轮箱的多种故障的振动信号,进行了样本熵特征提取,应用SFLA-PSO融合算法优化了BP神经网络模型,对行星齿轮箱故障进行识别诊断。结果表明:基于ANCEEMD样本熵特征提取的SFLA-PSO-BP诊断模型较PSO-BP和BP在行星齿轮箱故障诊断中的准确率分别提高了5%、15%。

基于LMD与相关分析的行星齿轮箱测点优化

针对机械传动设备行星齿轮箱极易发生故障的问题,减少行星齿轮箱故障诊断中传感器的布置数量,进而降低成本,在此提出一种基于局域均值分解(LMD)信息熵与相关分析相结合的行星齿轮箱测点优化方法。首先,使用LMD算法分解5种工况的振动测试信号;其次,用分解出的各分量与原始数据之间的相关系数筛选出包含主要故障信息的PF分量,并提取它们的信息熵特征,构造成样本特征向量;最后,对每种工况不同测点的信息熵特征向量进行相关性分析计算。统计分析各测点之间的相关性指标,对各测点进行筛选从而达到测点优化的目标。

Analysis on the Pressure Fluctuation Law of a Hydraulic Exciting System with a Wave-exciter

A hydraulic exciting system with a wave exciter has been constructed in order to study the hydraulic vibra- tion law. The system consists of an oil source, wave-exciter and oil cylinder, and is controlled by a wave-exciter. The working principle of the hydraulic exciting system and wave exciter has been analyzed, and its excitation process has been illustrated. The law of every pipe＇s pressure fluctuation of the system is obtained by experiment. The theo- retical analysis and experimental data prove that the pipeline pressure periodically changes and the pipeline pressure fluctuation frequency is independently controlled by the excitation frequency of the wave-exciter. Every pipelinc＇s pressure wave is produced by system flow fluctuation and water hammer coupling. The pressure fluctuation rules of the system provide a theoretical basis for the study of the associated liberation system.