Knowledge Update in a Knowledge-Based Dynamic Scheduling Decision System

  • Chao Wang
  • Zhen-Qiang Bao
  • Chang-Yi Li
  • Fang Yang
Part of the Lecture Notes in Computer Science book series (LNCS, volume 4092)


Through the interrelated concept of the job shop production, this paper constructs a dynamic scheduling decision system based on knowledge, and gives five attributes of resource agent and corresponding task, time, cost, quality, load and priority. Using the fuzzy set and rough set, the classified knowledge of the attribute is generated, and is used as the states criteria in the Q-learning. To initialize Q value of the decision attribute, we collect the knowledge from experts. The Q-learning algorithm and initial parameter values are presented in knowledge based scheduling decision model. By the algorithmic analysis, we demonstrate its convergence and credibility. Applying this algorithm, the system will update the knowledge itself continuously, and it will be more intelligent in the changeful environment, also it will avoid the subjectivity and invariance of the expert knowledge.


Reinforcement Learning Shop Floor Schedule Decision Resource Agent Schedule Rule 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Bao, Z.-q., Wang, N.-s., Cai, Z.-t.: Research on the model of scheduling agent based on bid using Rough-Fuzzy sets. Machine Engineering of China 14(22), 1943–1946 (2003)Google Scholar
  2. 2.
    Wang, Y.-C., Usher, J.M.: Learning policies for single machine job dispatching. Robotics and Computer-Integrated Manufacturing 20, 553–562 (2004)CrossRefGoogle Scholar
  3. 3.
    Emin Aydin, M., Öztemel, E.: Dynamic job-shop scheduling using reinforcement learning agents. Robotics and Autonomous System 33, 169–178 (2000)CrossRefGoogle Scholar
  4. 4.
    Gao, Y., Chen, S.-f.: Research on Reinforcement Learning Technology: A Review. Acta Automatice Sinica 30(1), 89–100 (2004)MathSciNetGoogle Scholar
  5. 5.
    Kong, L.-F., Wu, J.: Dynamic single machine scheduling using Q-learning Agent. In: Yeung, D.S., Liu, Z.-Q., Wang, X.-Z., Yan, H. (eds.) ICMLC 2005. LNCS, vol. 3930, pp. 3237–3241. Springer, Heidelberg (2006)Google Scholar
  6. 6.
    Huang, J., Yang, B., Liu, D.-y.: A distributed Q-learning algorithm for multi-agent team coordination. In: Yeung, D.S., Liu, Z.-Q., Wang, X.-Z., Yan, H. (eds.) ICMLC 2005. LNCS, vol. 3930, pp. 108–113. Springer, Heidelberg (2006)Google Scholar
  7. 7.
    Rabelo, L.C., Jones, A., Yih, Y.: Development of a real-time learning scheduling using reinforcement learning concepts. In: 1994 IEEE International Symposium on Intelligent Control, Columbus Ohio USA, pp. 291–296 (1994)Google Scholar
  8. 8.
    Bao, Z.-q., Li, C.-y., Zhou, X., Bian, w.-y.: A Knowledge-based Dynamic Scheduling Decision System. In: Proceedings of 2005 international conference on management science & engineering (12th), Incheon, R. Korea, pp. 1554–1559 (2005)Google Scholar
  9. 9.
    Sutton, R.S., Precup, D., Singh, S.: Between MDPs and semi-MDPs: a framework for temporal abstraction in reinforcement learning. Artificial Intelligence 112(1), 181–211 (1999)zbMATHCrossRefMathSciNetGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Chao Wang
    • 1
  • Zhen-Qiang Bao
    • 1
  • Chang-Yi Li
    • 1
  • Fang Yang
    • 1
  1. 1.Department of Computer Science and EngineeringYangzhou UniversityYangzhouChina

Personalised recommendations