AI-Based Prescriptive Analytics for Predictive Maintenance: A Kaplan–Meier and Machine Learning Approach

In the realm of industrial operations, ensuring the seamless operation of machinery is paramount. This paper presents an innovative approach to predictive maintenance that combines Kaplan–Meier survival analysis with machine learning techniques. The framework was tested using a synthetic dataset inspired by milling machine operations, focusing on addressing the crucial challenge of predicting machine failures and implementing proactive maintenance strategies. The model identifies machine failures resulting from five independent modes: tool wear failure (TWF), heat dissipation failure (HDF), power failure (PWF), overstrain failure (OSF), and random failures (RNF). Each mode presents unique challenges to industrial operations and maintenance. Our approach employs Kaplan–Meier survival analysis to estimate machinery survival probabilities under these failure modes. Subsequently, machine learning algorithms predict failure occurrences and offer prescriptive maintenance recommendations using Natural Language Processing (NLP). In various test scenarios, the model consistently achieves an impressive accuracy of 99.8%, demonstrating its robustness and reliability. The reason for implementing predictive maintenance lies in its ability to proactively identify and address potential machinery failures, thus minimizing downtime, optimizing resource allocation, and enhancing overall operational efficiency.