Enhanced process safety: Liquid level identification in the process of preparing PBXs by water suspension method using YOLOV8

Realizing online monitoring of the process of preparing Plastic Bonded Explosives (PBXs) by water suspension method has far-reaching significance for optimizing the preparation process, improving product quality, and ensuring personnel safety. The application of deep learning algorithms has significantly improved the efficiency of image recognition and is widely used in various industries. This study proposes a YOLOV8 framework based on deep learning to identify liquid level pictures during the preparation process of PBXs. The data sets are divided according to different split ratios, and each data set uses three levels of epoch of 100, 200 and 300 for training. Four key performance indicators, including precision, recall, mAP50 and mAP50-95, are used to evaluate the quality of the model. The results show that the model has outstanding recognition effect, and all indicators have reached more than 97%. This study innovatively applied the deep learning framework to the field of PBXs preparation, realized the monitoring of the PBXs preparation process, and provided new ideas and foundation for the advanced manufacturing of PBXs.