VIBRATION DATA IUT DOUALA

Each fault condition is further characterized by two severity levels (“speed 1” and “speed 2”), resulting in nine distinct states. Each condition contributes approximately 200,000 data points. The dataset is structured into four main variables: time (timestamp), vibration signal, fault type, and severity level.

For machine learning applications, the data are provided in CSV format and organized into training, validation, and test subsets, ensuring a balanced and representative distribution of all classes. To enhance realism and robustness, Gaussian white noise has been added to the signals, simulating industrial operating environments. The dataset is suitable for applications in fault diagnosis, condition monitoring, and predictive maintenance using machine learning and deep learning techniques.