Research on the Knowledge Acquirement and Updating in Dynamic Scheduling System

In a knowledge-based dynamic scheduling system, scheduling decision bases on historical experience knowledge, but along with the process of time and environmental change, the worm-eaten rule knowledge will affect the scheduling effect. This paper constructs a scheduling decision system based on the Contract Net Protocol. Through rough set, the system acquires scheduling rules knowledge from experience of experts. And the system examines whether the condition attribute combination matches up to the invitation of bid, and recalculates the degree of credibility of the resource Agent´s condition attributes. According to the degree of satisfaction which is evaluated from the done task´s result, the system obtains the set of the degree of satisfaction , and examines whether the decision attributes, one is in the knowledge base and the other is defined with the mean of the condition attribute combination, are match case. If they are not, the system updates the decision attributes. Then the system recalculates the degree of credibility of rule knowledge with the entropy of the set. Finally an example is presented and analyzed.