多属性决策的时间不确定事件流时序推理方法

针对信息物理融合系统时间不确定事件流调度顺序的决策依据单一问题,首先利用D-S证据理论在多证据源概率融合上的优势,充分考虑事件优先级、截止期、紧迫度、事件依赖等多个属性的影响,构建具备多属性特征的模糊结束时刻基本概率分配求解模型。然后,建立D-S证据理论与直觉模糊集的关联模型,求解模糊结束时刻隶属度与非隶属度;最后,利用直觉模糊集负向时间推理理论和相关计分函数推导模糊开始时刻概率得分,得到时序推理结果,并以此确定基于多属性判据的时间不确定事件流调度顺序。实验结果表明,当事件数量增长时,调度准确率可保持在85%以上;当模糊区间限制规模扩大时,调度准确率下降幅度不超过15%。

Method of uncertain temporal event flow sequence reasoning based on multiple attribute decision making

In the cyber-physical system, the scheduling sequence of uncertain temporal event flows has a single decision-making problem. To solve this problem, this paper firstly uses the advantage of D-S evidence theory in multi-evidence source probability fusion, considering the influence of multiple attributes such as event priority, deadline, urgency and event dependency, to construct a basic probability distribution model for fuzzy end time with multi-attribute features. Then, an association model with intuitionistic fuzzy sets and D-S evidence theory is established, and through this model, the membership and non-membership of the IFS at the fuzzy end time can be obtained. On this basis, negative temporal reasoning of IFS and correlation scoring function are used to derive the fuzzy start time probability score, and finally obtain the time series reasoning result, which is used to determine the uncertain temporal event flow scheduling order based on multi-attribute criterion. The experimental results show that when the number of events increases, the methods scheduling accuracy rate can keep more than 85%. And when the size of the fuzzy interval limit is expanded, the scheduling accuracy rate does not decrease more than 15%.