Multi-source heterogeneous data fusion using vector concatenation for quality fluctuation analysis

Big data mining is an important driver of quality control in intelligent manufacturing, however, conventional approaches often rely on a single data source, limiting their accuracy and applicability. In this study, a multi-source heterogeneous data integration method was developed for modeling quality fluctuations. One-dimensional image data are characterized using gray-level co-occurrence matrix parameters and fused with structured process data through vector concatenation. The fused features are then extracted using kernel principal component analysis to capture quality variations accurately. Experiments demonstrated a monitoring accuracy of 98.57% for the training set and 96.67% for the test set, outperforming models based on single-source data. From a managerial perspective, the proposed approach provides a data-driven tool for early detection of process anomalies, enabling production managers to reduce defects, improve operational efficiency, and support informed decision-making in intelligent manufacturing environments. This methodology advances both the precision and the applicability of industrial quality monitoring, offering practical value for data-driven production management.