Foreign Object Separation in Bulk Good Systems Using Machine Learning and Image Processing

One of the major goals in bulk good systems is to provide high-quality material. Sometimes the bulk good is corrupted by foreign objects, due to insufficient process observations, human mistakes, or unforeseen events. These particles are unwanted in the process. Furthermore, they can lead to unnecessary plant shutdowns or can even lock sub-parts of the plant. Current solutions work with filter techniques to eject foreign objects and ensure high quality. However, many solutions are difficult to implement due to process constraints and give no precise information about the corruption level and their objects. Therefore, a machine learning-based method is developed to detect objects in the bulk material and control an ejection system. In addition, based on the object position, the conveyor speed and air pressure injectors are controlled to separate the detected object during free fall. Consequently, the closed-loop structure and workflow for a custom air-pressure ejection process are presented. The framework is successfully applied in a modular bulk good system and provides adequate results in terms of object detection and object separation.