Predictive maintenance has become an essential component of smart manufacturing systems for reducing unexpected equipment failures and improving operational efficiency. This study presents a hybrid machine learning framework for predictive maintenance using sensor-based machine operational data. The dataset consists of 5,000 machine operating records collected from sensor monitoring systems, including parameters such as operating hours, temperature, vibration, pressure, humidity, and machine load. A Random Forest model is first applied to analyze feature importance and identify the most influential parameters associated with machine failure. Subsequently, an Artificial Neural Network (ANN) model is employed to predict machine failure conditions based on the selected operational features. The dataset is divided into training and testing subsets using an 80–20 split to evaluate the predictive capability of the model. Experimental results indicate that load percentage and operating hours are the most influential factors affecting machine failure, followed by vibration and pressure. The proposed framework achieved an overall prediction accuracy of 84.5%. Although failure events represent a small portion of the dataset, the developed system demonstrates the capability to identify potential machine degradation patterns and support data-driven maintenance planning. The results highlight the potential of hybrid machine learning approaches for improving predictive maintenance decision-making in smart manufacturing environments.

Keywords

Predictive maintenance, Machine learning, Smart manufacturing, Sensor data, and Failure prediction.