关键词: 数据驱动, 中药制药, 车间调度, 动态调度, 人工蜂群算法

Abstract: A dynamic scheduling problem of Chinese materia medica pharmaceutical workshop driven by multiple dynamic factors(DSP-CMMPW-MDF) model was established, the multiple dynamic factors such as raw material shortages, emergency order insertions, and machine breakdowns. An improved artificial bee colony with Q-learning(IABC-QL) algorithm was proposed to solve the DSP-CMMPW-MDF with the optimization objective of minimizing makespan. In the IABC-QL algorithm, an opposition-based learning strategy was proposed to generate the initial population, ensuring high quality and diversity of the population individuals. Five local search operations were designed to enhance the deep exploration capability of the algorithm. Thus the proposed model and algorithm were applied to a Chinese materia medica pharmaceutical granule production workshop. The results show that the proposed model may effectively improve the flexibility and adaptability of the production systems in the face of uncertainties. Additionally, a comparison with existing algorithms validates the effectiveness of the proposed algorithm.

Key words: data-driven, Chinese materia medica pharmaceutical, workshop scheduling, dynamic scheduling, artificial bee colony algorithm