The invention relates to the technical field of fault diagnosis, in particular to a tuning region fault diagnosis method based on multi-feature fusion and improved logistic regression, which comprises the following steps of: extracting time domain features from a fault signal by utilizing multi-time window analysis; a Welch algorithm is adopted to calculate a power spectrum and extract frequency domain features; fusing the time domain and frequency domain features by using an mRMR algorithm; and sending the fusion features into the improved classification model, and carrying out fault diagnosis. The method solves the problems of single fault feature type, to-be-improved feature characterization performance and high diagnosis algorithm cost during analysis of an existing method.

    本发明涉及故障诊断技术领域,尤其涉及一种多特征融合及改进逻辑回归的调谐区故障诊断方法,包括:利用多时窗分析对故障信号提取时域特征;采用Welch算法计算功率谱并提取频域特征;利用mRMR算法对时域、频域特征进行融合;将融合特征送入改进分类模型中,进行故障诊断。本发明解决现有方法分析时故障特征类型单一、特征表征性能有待提升、诊断算法代价高的问题。


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Tuning area fault diagnosis method based on multi-feature fusion and improved logistic regression


    Weitere Titelangaben:

    一种多特征融合及改进逻辑回归的调谐区故障诊断方法


    Beteiligte:
    FEI KELING (Autor:in) / CHEN SHUNZHI (Autor:in) / SUN XUMING (Autor:in)

    Erscheinungsdatum :

    2023-08-22


    Medientyp :

    Patent


    Format :

    Elektronische Ressource


    Sprache :

    Chinesisch


    Klassifikation :

    IPC:    G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / B61L Leiten des Eisenbahnverkehrs , GUIDING RAILWAY TRAFFIC / G01R Messen elektrischer Größen , MEASURING ELECTRIC VARIABLES