Diagnóstico de defeitos em unidades geradoras hidrelétricas com base na solução de problemas inversos

Abstract

The maintenance of rotating machinery is a topic of great interest in industrial sectors, with the main challenges being the early detection and diagnosis of mechanical failures in these systems. In this context, the research field dedicated to monitoring and diagnosing faults in rotating equipment aims to ensure operational efficiency and safety. The most commonly observed mechanical faults in the context of rotating machinery are mechanical unbalance, misalignment, and cracks. Various approaches can be adopted to model and detect these faults. This work developed a fault detection tool based on solving an inverse problem, which combines a mathematical model with the Differential Evolution optimization algorithm. The approach is based on a numerical model of the system, in which numerical vibration amplitudes are compared with experimental data obtained from the actual machine. New fault scenarios are introduced into the mathematical model, generating new vibratory responses. An iterative optimization process is then applied to minimize the difference between numerical and experimental results until both vibratory responses are similar. This approach enables the identification and characterization of the type and severity of the fault. The methodology was applied to the numerical model of one of the generating units of the Foz do Chapecó Hydroelectric Power Plant. The tool was used in two stages. The first stage focused on the unit's calibration, determining the residual unbalance of the turbine and the generator. This stage creates a digital twin of the machine, which is used in the diagnostic process. Then, with the machine calibrated, the second stage was applied, in which misalignments and cracks were introduced into the system to validate the proposed methodology. The results obtained with the application of the diagnostic methods showed relative errors below 1,36%, confirming the feasibility of the adopted approach.