System production rate evaluation of multi-process serial system based on improved cross-merging method

• Published in: 2023 5th International Conference on System Reliability and Safety Engineering (SRSE) pp. 458 - 462
• Main Authors: Bai, JinZhuo, Wang, JingCheng, Duan, WeiWei, Dai, Wei
• Format: Conference Proceeding
• Language: English
• Published: IEEE 20.10.2023
• Subjects: Adaptation models; Analytical models; improved cross-merging simplified method; Merging; Monte Carlo methods; multi-process serial system; production circulation model; Productivity; Steady-state; System performance
• DOI: 10.1109/SRSE59585.2023.10336129

As the integration and informatization of enterprise manufacturing processes continue to increase, manufacturing systems are evolving towards more complex structures, raising higher demands for the analysis methods of multi-process serial system performance. Meanwhile, existing evaluation methods face high computational complexity and poor applicability when modeling and analyzing large-scale, highly complex systems, making it difficult to provide scientifically reliable optimization strategies. In order to easily analyse complex multi-process serial system, this paper proposes an improved cross-merging simplified method to analyse the performance of multi-process serial system with whole process. Firstly, the production circulation model is established based on machines and buffer zones. Secondly, the multi-process serial system is simplified to a two-process model based on the proposed method and its performance is analyzed. The proposed method cross-uses the equivalent machine of forward and backward merging, which can comprehensively consider the influence of the upstream and downstream parts of the buffer zone. Time parameters are added in the iteration process, which can dynamically display the change curve of performance productivity of the multi-process serial system from transient to steady-state stage. Comparing to the results of 10,000 Monte Carlo simulation processes, the result has a high degree of accuracy. The performance evaluation error under different lengths process is calculated, and the average error can be guaranteed to be within 0.6%, indicating that the proposed method can maintain high accuracy and adaptability when evaluating the performance of multi-process serial systems at all stages.