Predictive maintenance has gained popularity in intelligent and sustainable manufacturing because it may increase uptime, eliminate unscheduled downtime, and maximize resource use. This research proposes an improved AdaBoost algorithm and big data analytics for intelligent manufacturing system predictive maintenance. First, the proposed system gathers and preprocesses massive industrial data from sensors, IoT devices, and other sources. The meta-algorithm AdaBoost improves mediocre learners' performance, predicting machinery failure and degradation. Algorithmic optimization adjusts hyperparameters like iterations and learning rate to balance model accuracy with processing efficiency. The proposed model improves on previous work with 0.972 accuracy, 0.977 precision, 0.972 recall, and 0.974 F1-score. An optimized AdaBoost-enabled predictive maintenance framework manages big data analytics complexity in manufacturing in a scalable, cost-effective, and smart way. The framework uses AdaBoost's potential and optimization approaches to improve predictive maintenance tactics, intelligent manufacturing, and industrial sustainability.