EHDE和WHO-SVM模型在齿轮箱故障诊断中的应用

Application of EHDE and WHO-SVM model in gearbox fault diagnosis

针对现有齿轮箱故障诊断方法对数据长度敏感的缺陷,提出了一种基于增强层次多样性熵(EHDE)和野马算法(WHO)优化支持向量机(SVM)的齿轮箱故障诊断模型。首先,传统熵值特征提取方法在特征提取阶段对数据样本的长度比较敏感,为此提出了增强层次多样性熵,并将其作为特征提取指标用于提取齿轮箱的故障特征;其次,采用WHO算法对SVM模型的参数进行了优化,建立了参数最优的WHO-SVM分类器;最后,将故障特征样本输入至WHO-SVM分类器中进行了训练和识别,完成了样本的故障识别;利用齿轮箱数据集分别从数据长度敏感性、算法特征提取时间、模型诊断性能三种角度对EHDE、精细复合多尺度样本熵、精细复合多尺度模糊熵、精细复合多尺度排列熵、精细复合多尺度散布熵、精细复合多尺度波动散布熵进行了对比研究。研究结果表明:EHDE方法对数据长度的要求较低,在数据长度为512时即可以取得99.1%的平均识别准确率,在诊断稳定性和诊断精度方面均优于其他对比方法;在算法的泛化性实验中,EHDE方法能够以98%的准确率识别齿轮箱的不同故障类型,具有明显的泛化性和通用性。

Aiming at the defect that the existing gear box fault diagnosis methods were sensitive to data length,a gearbox fault diagnosis model based on enhanced hierarchical diversity entropy(EHDE)and wild horse optimizer(WHO)optimized support vector machine(SVM)was proposed.Firstly,the traditional entropy value feature extraction method was sensitive to the length of the data sample in the feature extraction stage,so the enhanced hierarchical diversity entropy method was proposed and used as a feature extraction index to extract the fault features of the gearbox.Secondly,the WHO algorithm was used to optimize the parameters of SVM model,and a WHO-SVM classifier with optimal parameters was established.Finally,the fault feature samples were input to WHO-SVM classifierfor training and test,and the fault identification of the samples was completed.By using gearbox data set,EHDE,refined composite multiscale sample entropy,refined composite multiscale fuzzy entropy,refined composite multiscale permutation entropy,refined composite multiscale dispersion entropy and refined composite multiscale fluctuation dispersion entropy were compared from three perspectives:data length sensitivity,algorithm feature extraction time and model diagnosis performance,respectively.The research results show that EHDE method has low requirements on data length,and can achieve 99.1%average recognition accuracy when the data length is 512,which is superior to other comparison methods in terms of diagnostic stability and diagnostic accuracy.In the generalization experiment of the algorithm,EHDE method can identify the different fault type of the gearbox with 98%accuracy,which has obvious generalization and universality.