This thesis investigates anomaly detection and classification in a simulated modular manufacturingenvironment using Machine Learning algorithm Random Forest. This algorithm is tested on a localcomputer and an embedded device, specifically the Raspberry PI. The performance of Random Forestmodels is evaluated for anomaly detection and classification tasks, considering different evaluationmetrics and execution time. The results indicate variations in model performance across differentmodules and classification tasks. It is observed that the limited computing resources of the RaspberryPI for anomaly detection tasks lead to significantly higher prediction times compared to a computer,highlighting the impact of embedded systems’ constraints on ML model execution