中文摘要：

针对基于知识的调度系统在实际应用中存在的知识源具有主观倾向性、决策依赖部分属性等缺陷，提出了一种新的基于知识的动态调度决策方法。在此基础上设计了调度决策系统。以合同网协议为协调机制，构建了基于多Agent的动态调度框架；由调度专家经验知识获取调度规则知识，并采用投票表决法解决规则知识的冲突，在实现知识融合的同时获得规则知识的可信度；对资源的调度历史记录进行学习，获取资源的条件属性可信度；利用粗糙集中定义的奈件属性重要度，计算调度规则知识中各条件属性重要度，并对其进行归一化处理，获取条件属性权值；提出了基于模糊Petri网的知识决策模型，以条件属性可信度为托肯、条件属性权值为输入强度、规则知识可信度为输出强度进行推理，得到决策属性可信度，形成决策结果。描述了动态调度系统的知识决策过程，完成了系统分析设计。

英文摘要：

Through reuse the acquired knowledge, knowledge--based scheduling reduces the load of manpower and achieves the agility of decision. Nevertheless, there are some constraints during the application of most systems as below： First, the sources of knowledge have their own subjective tendentiousness; second, scheduling decision making relies on partial attributes. This paper presents a knowledge-based decision system for dynamic scheduling. We construct the structure of dynamic scheduling on the basis of multi-agent theory and with Contract Net Protocol as the coordinating mechanism. The system acquires standardized attributes knowledge from bidding documents and knowledge of scheduling rules from experience of scheduling experts. By means of balloting, it solves the conflict among scheduling ＇rules to achieve fusion of knowledge. Furthermore, the degree of each rule＇s credibility is got from supporting rate. Through learning the records of scheduled resources, the system acquires the degrees of credibility of each resource＇s attributes. The degree of each attribute＇s importance in the knowledge of scheduling rules is figured out according to its definition in Rough Sets, and then standardized to acquire the weight of each attribute. The system contains a Knowledge-Based Decision Making Model which is on the basis of Fuzzy Petri Net theory. So we take the degrees of attributes＇ credibility for tokens, the weight of each attribute for the intensity of input and the degree of scheduling rule＇s credibility for the intensity of output to reason. As a result, the degree of decision attribute＇s credibility is generated. On the basis of above theories, we describe the process of knowledge-based decision making in the system and accomplish the design of its structure.