In our research, we will be working on abnormal detection in industrial ovens, which is essential in keeping the ovens in a state to run safely and efficiently. We proposed a novel method by combining real-time data from the IoT sensor installed in the Marel Modular Oven System (MOS) 700H and the XGBoost machine learning algorithm. Our method targets explicitly detecting leaking steam valves, a common fault that can lead to inconsistent product quality and quantity produced. We use XGBoost, a very efficient machine learning method, to handle temperature, humidity, and air pressure data. This allows us to detect anomalies in real-time accurately. The efficiency of the suggested model was thoroughly experimented with by training using real-world data collected over various periods. The results demonstrate our model’s superior performance to machine learning and state-of-the-art (SOTA) deep learning models. These findings demonstrate XGBoost’s ability to detect anomalies in industrial environments, especially its ability to manage imbalanced datasets and complex temporal dependencies. This study promotes effective monitoring and abnormal detection systems for industrial ovens, promoting improved product consistency and productivity in the food processing industry.