Purpose. Induction motors (IM) are the most important components in commercially available equipment and industrial processes. In this work, a framework based on the relevance vector machine (RVM) and generalized singular value decomposition/linear discriminant analysis (GSVD/LDA) used as fault detection of broken rotor bars (BRB) in IM is presented. We have obtained some important experimental results, which may become very helpful for the fault detection in induction motors. Methodology. First, we used GSVD/LDA to reduce the dimension of stator current. Then the multiclass classifier as the fault diagnosis was decomposed into several binary RVM classifiers using One-vs-All method and their kernel parameters were determined through cross-validation. Findings. By reducing the dimension of the current signal through GSVD/LDA, we lowered its redundancy significantly, so that the several binary RVM classifiers using One-vs-All method can perform better. The simulation results demonstrated that, the proposed method is more effective in identifying the fault of BRB in IM as compared with its counterpart. Originality. We made a study of fault detection of broken rotor bars in induction motor by combining GSVD/LDA-based dimension reduction with the binary RVM classifiers using One-vs-All method. We verified its effectiveness through the simulation. We have not found previously-made researches on this aspect. Practical value. The proposed method can be embedded in the fault diagnosis system of IM as a separate module. Since it shows the satisfactory accuracy, it can provide a reliable anchor for industrial and agricultural productions.

  Якщо, ви не знайшли інформацію про автора(ів) публікації, маєте бажання виправити або відобразити більш докладну інформацію про науковців України запрошуємо заповнити "Анкету науковця"