The dataset was generated using a turbine model by introducing wear faults, ovalization, and a combination of ovalization with wear in the hydrodynamic bearings, as described in the paper available at https://doi.org/10.1016/j.mechmachtheory.2024.105819. This dataset contains two types of data. The first comprises information from bearing simulations, including the equivalent stiffness, and damping coefficients as well as the shaft eccentricity values. The second type consists of input and output matrices for the neural network. The input matrices represent attributes of the dynamic response, while the output matrices correspond to the discretized non-circular bearing profiles.