Key Process and Quality Characteristic Identification for Manufacturing Systems using Dynamic Weighting Function and D-S Evidence Theory

摘要

Monitoring and controlling of process and quality characteristic for manufacturing system is a key issue in closed-loop quality control. Meanwhile, it is challenging due to the modern manufacturing system generally consists of hundreds of processes that are aligned to produce a specific end product. And the output quality of each individual process may be judged by dimensional quality characteristics. Each process and quality characteristic being monitored is costly and impractical. This paper attempts to provide a systematic approach to identify the key processes and quality characteristics simultaneously. Firstly, a modified casual matrix is used to acquire the correlation data between process and characteristic. The correlation degree is weighted by dynamic weighting function based on the importance of quality characteristics. Then, the triangular fuzzy function is used to construct the frame of discernment based on single index (quality characteristic). The mass functions that represent the degree of belief supported are determined and treated as pieces of evidence. Afterward, all of the evidence are combined by D-S (Dempster-Shafer) fusion rules. In addition, key quality characteristics are also identified based on the cumulative sum of the weighted score and Pareto Principle simultaneously. Finally, the usefulness of proposed approach is verified by a real-time dense medium coal preparation case.