Desenvolvimento de uma técnica de detecção de curto-circuito entre espiras do estator para aplicação em conversores de frequência com controle orientado de campo

Abstract

Fault Detection (FD) in its incipient stage is crucial for the prevention of unscheduled interruptions and to ensure the availability and reliability of production processes. With the advancement of condition-monitoring methods for devices, accurate evidence-based decisions are more tenable, while enhancing predictive maintenance strategies and asset management, which in turn increase equipment availability and industrial process profitability. However, even though certain critical industrial applications employ induction motors controlled in a closed-loop, this category of drives receives less attention in the literature concerning FD. Additionally, fault diagnosis in its early stages is even more challenging in such systems than in motors connected directly to the grid or open-loop inverters. In this context, this work explores the mechanisms influencing fault signatures of Inter-Turn Short-Circuits (ITSC) that hinder the diagnosis of such short-circuits, as well as presenting a detailed and comprehensive review of the problem. Furthermore, this work aims to develop a non-invasive methodology for the FD of incipient ITSC in motors driven by Field-Oriented Control (FOC), as well as to enable monitoring of the fault severity evolution. Among the contributions of this study is the proposition of a detection index based on the energy concept of the Discrete Wavelet Transform (DWT) and a method for selecting the respective decomposition levels to be used in the composition of the fault detection index. The proposed detection methodology utilizes motor current signals that have already been measured and employed by the control structure. The results, when compared to those derived from an existing method in the literature, demonstrate that the developed technique is effective for motors driven by FOC, robust to noise and control reaction mechanisms, and computationally inexpensive. This makes it attractive for implementation in frequency converters. This work thus contributes to the fields of fault diagnostics, condition monitoring of equipment, reliability, and asset management.