融合粗糙集和证据理论的温室环境控制推理决策方法

针对温室环境控制专家决策的需求,该文提出了一种融合粗糙集和证据理论的推理方法,建立基于粗糙集和证据理论的决策模型,模型推理流程包括连续属性离散化、专家决策表形成、属性约简、证据组合推理。首先采用模糊C均值聚类方法离散连续的环境指标数据,利用基于信息熵的属性约简算法对决策表进行优化,剔除专家知识中的冗余成分,然后引入证据理论组合优化后的指标,最后依据基于基本可信度分配的决策,判定温室应采取的控制方法。实例研究表明该方法能有效提高控制决策可信度,减小决策的不确定性,将其应用于温室环境控制决策具有可行性。

Decision-making method based on rough set and evidential theory for greenhouse environmental control

Aiming at the requirements of expert decision-making for greenhouse environmental control, an inference method based on rough set and evidential theory was proposed. The decision-making model includes continuous variables discretization, formation of expert decision-making table, attribute reduction and evidence combination. The decision-making model was established by four steps. Firstly, discrete the consecutive environment index data through fuzzy C means clustering method. Secondly, optimize decision table by using attribute reduction algorithm based on information entropy, so that to eliminate the redundancies of expert knowledge. Thirdly, introduce the theory of evidence to process the optimized index. Finally, judge an appropriate greenhouse control method according to basic probability distribution decision. The case study indicates that decision making method can greatly enhance the reliability of decision-making and reduce its inference uncertainty, which has an important significance to the application in greenhouse environmental control.