Complex product quality diagnosis framework based on association rule mining

Quality diagnosis serves as a critical tool in quality control, playing a pivotal role in identifying potential quality issues and thereby enhancing the overall quality management of enterprise products. However, conventional quality diagnosis methods often underutilize the valuable data relationships between process parameters and quality indicators, resulting in challenges in providing a clear diagnostic foundation for quality anomalies, thereby complicating quality control efforts. To address this issue within the context of modern manufacturing data characteristics, this study proposes a product quality diagnosis method based on association rule mining. The methodology begins with the application of the Partitioning Around Medoid (PAM) algorithm for data pre-processing, including clustering and discrete coding. Subsequently, the Frequent Pattern Growth (FP-Growth) algorithm is employed to mine rules from the discretely coded data. Finally, the Whale Optimization Algorithm (WOA) is leveraged to extract optimal rules and establish a comprehensive rule library for effectively tracing the root causes of anomalies, thereby enabling precise quality diagnosis of products.