Electrical signals dataset from fixed-speed and variable-speed synchronous generators under healthy and faulty...

Proper monitoring of rotating machines is responsible for the efficiency in detecting, diagnosing, or even prognosing failures. Effective monitoring can lead to increased economic viability of any equipment, as it may reduce costly repairs, decrease downtime, and increase safety. Knowing the behaviour of a machine promotes better monitoring of its operation and maintenance. Data-driven algorithms have been widely used to identify failures and predict the behaviour of machines and systems. The difficulty in obtaining reliable data to test strategies or methods for this purpose is well known. Our contribution is a set of electrical current data (time series data) from a rotating machine that generates electrical energy, generically called a power generator, in a laboratory. In this machine we have the possibility of, besides monitoring its healthy behaviour, causing internal defects that can reduce its efficiency and remaining useful life. We highlight three key lines of study with the data available here: it is possible to apply data processing tools to make discoveries not evidenced in studies; test and compare new data-driven algorithms using public and reliable data; engineering lectures can use the dataset regarding the study of electrical machines and data driving methods. The dataset contains information mainly about the voltage and current of generators when they are subject to internal faults. These faults include short circuits between turns of winding, short circuits between windings of the same phase, and short circuits between different phases. This dataset has a wide variety of bench configurations. The dataset comes from real generators and allows the study of phenomena that are difficult to reproduce through analytical or computational models. The time series of electrical currents are raw, no preprocessing has been done. In fact, the signals contain natural noise from an industrial environment. In this context, the main contribution of this work is to provide a public and reliable database, which helps to speed up the development of more efficient techniques for monitoring, diagnosis, and prognostics of the behaviour of rotating electrical machines.