基于 EMD-EDT 的行星齿轮箱特征提取及状态识别方法研究

摘要

Aiming at the difficulties in planetary gearbox feature extraction and condition recognition,36 suitable feature parameters for planetary gearbox condition recognition were summarized and the originating and extraction method of each parameter were introduced.A two-stage characteristic parameters selection and weighting method based on EDT were proposed to get a new composed characteristic parameter.An IMF selection method was proposed based on the energy ratio.And then a planetary gearbox condition recognition method was put forward based on EMD and EDT. Extracting the new composed feature parameter of each selected IMF and constituting the characteristic matrix as the input in EDT calculation,the evaluation results were output intelligently.The data of experiments with preset planetary gearbox fault were utilized to validate the effectiveness of the proposed method and the necessity of EMD,feature selection and weighting.%针对行星齿轮箱特征提取及状态识别中存在的难点问题，首先总结适用于行星齿轮箱状态识别的36个特征参数并详细说明各参数的来源及提取方法，提出基于 EDT 的两阶段特征参数选择及加权方法，得到新的组合特征参数。提出基于 EMD-EDT 的行星齿轮箱状态识别方法；首先提出基于能量比的包含敏感故障信息的固有模态函数（IMF）选择方法，然后选择提取筛选出的各 IMF 的组合特征参数构成特征矩阵作为 EDT 的计算输入，最后智能的输出评估结果。利用行星齿轮箱预置故障实验数据验证所提出方法的可行性和有效性及 EMD 分解、特征选择和加权过程的必要性。