The vertical roller mill used in the cement industry is a large piece of equipment that works without interruption and is essential in production. Due to the occurrence of several breakdowns of its components, mainly the roller ends that caused several production stops, the idea of analyzing the vibrations of the mill to identify a possible defect pattern arose. Sound analysis was performed during a period of terminal failures. The application of signal processing techniques such as FFT, multiresolution analysis, and SAC-DM (chaos theory) allowed the observation of a pattern of equipment behavior that preceded the appearance of failure in the vertical mill's roller terminals. It was observed that the multiresolution analysis and the SAC-DM were sensitive to these patterns, allowing the company's maintenance sector to prepare for repairs.