Anomaly Detection in Industrial IoT Sensor Data in Matlab

The rapid adoption of the Internet of Things (IoT) in industrial environments has transformed conventional monitoring systems into intelligent and data-driven infrastructures. Industrial IoT (IIoT) sensors continuously generate large volumes of operational data, including temperature, pressure, vibration, and humidity measurements, enabling real-time supervision of industrial equipment and processes. However, manually identifying abnormal behavior within such high- dimensional and continuously streaming data is both time-consuming and inefficient. Therefore, the integration of machine learning techniques for automated anomaly detection has become essential for ensuring operational reliability and minimizing equipment failures. This research presents a comparative analysis of two supervised machine learning algorithms—Support Vector Classifier (SVC) and Random Forest—for anomaly detection in Industrial IoT sensor data using the MATLAB environment. Initially, the collected sensor dataset undergoes preprocessing procedures such as data cleaning, handling missing values, normalization, and feature scaling to improve model performance and reliability. The processed dataset is subsequently divided into training and testing subsets for effective model evaluation. The Support Vector Classifier model utilizes a Radial Basis Function (RBF) kernel to identify complex nonlinear relationships within sensor data and classify operational states as either normal or anomalous. In parallel, the Random Forest algorithm employs an ensemble learning strategy by integrating multiple decision trees to enhance classification accuracy and improve robustness against noisy and inconsistent sensor readings. The performance of both models is assessed using evaluation metrics including accuracy score, confusion matrix, precision, recall, F1-score, and Receiver Operating Characteristic (ROC) curve analysis. Experimental findings indicate that both machine learning models effectively detect anomalies in industrial sensor datasets. However, the Random Forest classifier demonstrates comparatively superior performance due to its enhanced generalization capability, resistance to overfitting, and efficient handling of multidimensional data. The proposed approach highlights the practical significance of machine learning-based anomaly detection in Industrial IoT systems for predictive maintenance, fault diagnosis, operational safety, and reduction of unplanned industrial downtime. This study confirms that intelligent anomaly detection frameworks can substantially improve the efficiency, reliability, and sustainability of modern industrial operations