最大熵不确定型决策模型与应用

Maximum entropy methods for uncertainty decision-making

现有不确定型决策方法中,各种状态的不确定性没有在决策方法中得到体现,且没有充分挖掘利用益损值信息,不同的决策准则往往得到不同的决策结果,不利于决策者做出正确决策.鉴于此,提出了基于信息不确定性的最大熵决策模型,通过构建状态分布的信息熵和决策结果离差信息的多准则目标函数,以状态概率分布为优化变量,采用加速遗传算法优化求解.防洪调度决策分析结果显示,最大熵决策模型可充分挖掘益损值信息,基于信息熵的不确定性准则使状态概率分布更加合理,决策结果可区分性优于其他决策方法.

The problem of existing uncertainty decision-making methods is that the uncertainty ot vartous states was not reflected, and the data information of gain/loss value was not fully mined. Sometimes, different decision criteria lead to a different decision results,which would confuse the decision makers for the fight decisions. In view of this, the maximum entropy decision-making method is proposed based on the information entropy of various states and the deviation information of decision-making results, taking the state probability distribution as optimization variables,and accelerating genetic algorithm used to optimize the multi criterion obiective function. The flood dispatching decision analysis shows that the maximum entropy model can fully mine the gain/loss information,and the state probability distribution is more reasonable,and the distinctiveness of deci-sion-making results is superior to that of other decision methods.