Abstract: To select the optimal construction solution of the intelligent factory,an evaluation-criteria system was designed from three support dimensions of intelligence,production and marketing and auxiliary.Considering interval-number criterion weights and normal cloud criterion evaluations,a novel method was proposed based on deviation-consistency principle and cosine approximations.According to the relation between normal clouds and normal stochastic variables,a normal cloud decision matrix was transformed into a generalized normal stochastic decision matrix.Through the 3σ principle,an interval-number decision matrix was generated.Furthermore,the rationality and feasibility of interval-criterion weight vector were tested.Using the entropy-weight technique and deviation-consistency principle,an integrated criterion-weight vector was gotten with a nonlinear multi-object programming model.Based on a normalization interval-number decision matrix,considering the differences of directions and positions between alternatives and ideal alternatives,the improved weighted interval-cosine similarities and approximations were attained.Accordingly,the ranking order was gained.By taking an agricultural equipment-manufacturing enterprise as the example,the feasibility and validity of the proposed method were verified.

Key words: intelligent factory, construction solution, multi-criterion optimal selection, normal cloud, deviation-consistency principle, cosine approximation, agricultural equipment-manufacturing enterprise

摘要： 针对模糊随机环境下智能工厂建设方案优选问题,从智能、产销、辅助3个支持维度设计评估准则体系,考虑区间数准则权重与正态云准则值,提出基于偏差一致性原理与余弦逼近度的新方法。首先,根据正态云与正态随机变量间的关系,将正态云决策矩阵转化为广义正态随机决策矩阵,通过正态云的3σ原则生成区间数决策矩阵;然后,检验区间准则权重向量的合理性和可行性,运用熵权法与偏差一致性原理构建非线性多目标规划模型,求取综合准则权重向量;进一步,基于规范化区间数决策矩阵,考虑各方案与理想方案的方向与位置差异,计算改进型加权区间余弦相似度及加权区间余弦逼近度,获得各方案的排序;最后,以农业装备制造企业为例,验证了所提方法的可行性和有效性。

关键词: 智能工厂, 建设方案, 多准则优选, 正态云, 偏差一致性原理, 余弦逼近度, 农业装备制造企业