Quality Control Model for Robotic Filling Lines via Fault Localization

The article proposes a mathematical model aimed at improving the operational quality of robotic filling lines through the detection and localization of faults in key production operations such as dispensing, capping, labeling, and packaging. The operational quality of robotic filling lines in this context is defined by the intensity of deviations from the normal state of an operation, including the probability of Type I errors (false positives) and Type II errors (false negatives). The primary objective of the research is to develop a model that enables timely detection of deviations from normal operation and precise identification of fault locations, thereby maintaining the desired operational quality of the production line. The proposed model describes the production process as a sequence of key operations, each characterized by probabilistic parameters, such as the frequency of function calls and the temporal execution parameters. Deviation metrics are calculated to assess the degree of non-conformity of actual operation performance with acceptable norms. Simulation modeling based on real production system data demonstrated that the model not only detects anomalies but also localizes them at the level of specific operations. Accuracy evaluation of the model revealed that it achieves high specificity (99.5%) and sensitivity (96%) in identifying anomalous cycles. This indicates the model's ability to correctly detect abnormal conditions (e.g., deviations in dispensing or capping), while its high specificity reflects a low likelihood of incorrectly classifying a normal operational cycle as anomalous. Thus, the proposed model serves as a universal tool for quality management of automated production lines, applicable to various industrial sectors, and contributes to the further development of quality control methods for production operations in automated lines.