In the context of Industry 4.0, predictive maintenance is emerging as a highly innovative and efficient solution to enhance the reliability, availability, and overall performance of plastic injection molding machines. These machines depend heavily on three-phase asynchronous motors, which play a critical role in the transformation and shaping process of plastic materials. However, these motors are susceptible to abnormal temperature increases, which can significantly compromise product quality, reduce operational efficiency, and lead to unexpected and costly equipment failures. Overheating may result from various factors, including excessive mechanical load, insufficient or inefficient cooling systems, electrical imbalances or faults, as well as challenging environmental conditions such as high ambient temperatures and poor ventilation. To address these issues, this work proposes a comprehensive methodology that leverages machine learning algorithms and real-time sensor data to monitor and predict the thermal behavior of induction motors with an integrated cooling system. By analyzing large volumes of data collected from sensors, predictive models can detect early signs of thermal anomalies and potential failures. This allows for more accurate and timely maintenance interventions, reducing unplanned downtime, extending equipment lifespan, and optimizing maintenance resources. Ultimately, the integration of artificial intelligence into maintenance practices enables a proactive, data-driven strategy that shifts from reactive interventions to predictive, performance-oriented maintenance in smart manufacturing environments.