【摘要】 针对管理实践及大数据处理过程中具有多决策属性的粗糙集属性约减问题,将条件属性依赖度与知识分辨度进行结合构建属性权重,分别建立针对不同决策属性的约减目标函数,引入帕累托最优思想,将基于多决策属性的粗糙集属性约减问题转化为离散多目标优化问题。针对该问题的结构设计了具有集群智能优化思想的元胞自动机求解算法,在算法中引入基于个体的非支配解集平衡局部最优与全局最优的关系,引入混沌遗传算子增加种群多样性。以某铁路局设备安全风险处理数据为案例构建多决策属性粗糙集决策表进行优化计算并进行管理决策分析。研究发现:(1)相对于传统的NSGA-II与MO-cell算法,本文提出的算法具有更强的多目标属性挖掘性能;(2)帕累托最优思想可以较好地解释多决策属性粗糙集在管理实践中的意义。更多还原

【Abstract】 Fortheproblemofattribute reduction ofroughsetwithmultidecisionattributesinthemanagementand big data processing, objective functions for differentdecisionattributesareestablishedbasedonthe dependability and knowledge resolution of condition attributes. Byintroducingtheideaof Pareto optimal solution, theattribute reduction problemofroughsetwithmultidecisionattributes is converted into multi-objective optimization problem with discrete variables. Basedontheideaof swarm intelligence, the new cellular automata solutionalgorithmisdesigned, and in the algorithm, the set of individual non-dominated solutions isdesignedtobalancethe local optimum andtheglobal optimum.Moreover, chaoticoperatorisintroducedtoincreasethe population diversity. Inthe numerical example, the decision tablesareestablished, optimizedandanalyzedbasedonthesecurity risk handling dataofarailway company. The studyfindsthat(1)comparedwiththetraditionalNSGA-II andMO-cellalgorithms, the new algorithmshowsbetterperformancein multi-objective attribute mining;(2)theideaof Pareto optimal solution can well explaintheapplicationofroughsetwithmultidecisionattributesinmanagement.更多还原