Desenvolvimento de métodos de detecção de falhas para o problema de barra de rotor quebrada utilizando análise baseada em modelos, análise do espectro de corrente e avaliação de potência

Abstract

This work presents contributions for Induction Motor (IM) Broken Rotor Bar (BRB) fault detection in three different approaches, namely Model-Based Fault Detection (MBFD), Motor Current Signature Analysis (MCSA), and Power Quantities Fault Detection (PQFD). The MBFD contributions are considered where fault detection is performed by combining Linear Parameter Varying (LPV) observers with the Virtual Rotor Current (VRC) modeling. The method is validated using simulated and experimental data for steady-state line-connected IMs, inverter-fed IMs, both for incipient and severe fault conditions under different load configurations and voltage unbalance, demonstrating its ability to detect faults in a wide operational range. The MCSA contributions are developed by the use of odd triple harmonics in delta-connected IMs, components that arise around the main odd triple components carrying information about the IM condition. The mathematical modeling introducing the harmonic components, along with experimental results under several load levels, voltage unbalanced, and under inverter-fed power supply, demonstrates the effectiveness of the detection. Finally, the contributions on the PQFD are achieved by using Conservative Power Theory (CPT) to identify BRB faults. In the proposed method, statistical features are extracted from the CPT power quantities, evaluated using classification tools. The treatment of incipient fault severity is also addressed by assessing the likelihood of a motor being healthy or faulty.