Study on fault detection and new monitoring techniques for im based on MCSA improvements

  1. Cusido Roura, Jordi
Dirigida por:
  1. Juan Antonio Ortega Redondo Director/a
  2. José Luis Romeral Martínez Director/a

Universidad de defensa: Universitat Politècnica de Catalunya (UPC)

Fecha de defensa: 28 de marzo de 2008

Tribunal:
  1. Joan Peracaula Presidente/a
  2. Jordi-Roger Riba Ruiz Secretario/a
  3. Manés Fernández Cabanas Vocal
  4. Martha Cecilia Amaya Enciso Vocal
  5. Jordi Catalá López Vocal

Tipo: Tesis

Teseo: 145726 DIALNET

Resumen

Diagnosis of industrial applications is one of the workhorses for researchers around the world since fault conditions imply great costs resulting from the repair of motors and the stopping of industrial processes and production lines, These problems need to be solved, or at least minimized. Many efforts have been made to better understand fault phenomena in Induction Machines. However, the discipline of fault detection requires electrical, mechanical, electronic and signal-processing knowledge. For the proper detection of faults, different and complementary knowledge is needed. Not only is advanced knowledge of electrical machines involved in this problem, but also signal-processing methods. Our main goal is to go further into these disciplines to increase our knowledge and make it possible to define various diagnostic solutions for every motor operation position. This research evolves around the MCSA (Motor Current Signature Analysis) diagnosis method, which is based on stator-current condition monitoring. This thesis attempts to analyze several disciplines in order to properly understand fault behaviour and this effect on Induction Machines. In support, theoretical analysis of this problem and also machine modelling are proposed. In order for these generic goals to be accomplished different and specific objectives must be defined. Those objectives are the following: ¿ Development of a Parametric Model for Faults based on: - Parameter-variation evaluation based on FEM tools - Simplified parametric model ¿ Study of a Specific Diagnosis Protocol based on test signal injection. ¿ Improvements in classical methods for motor diagnosis based on time-frequency transforms by means of: - Short Time Fourier Transform - Evaluation of new strategies for fault detection based on advanced signal-processing techniques, such as the Wavelet Transform - Development of new techniques of signal processing based on convolutions and Wavelet functions - Intro