This paper presents a novel Soft Computing Approach called “Neuro-Fuzzy Analytical Network Process (NFANP)” for the group decision-making problems based on the conventional Analytic Network Process (ANP) method. The proposed approach deals with the interval values of judgments in a fuzzy environment using mobile, not fixed, trapezoidal and triangular membership functions, as well as the interval numerical ratio defined by alpha-cuts and the decision maker’s confidence levels. The consistency problem of the fuzzy reciprocal matrices is addressed in the proposed paper by allowing a certain tolerance deviation to be less than 0.20. Furthermore, trained Artificial Neural Networks (ANNs) are included in the proposed approach to reduce the large number of computations of the arithmetic operations required to correlate decision factors with the alternatives. In the proposed implementation, the selection problem is defined into three main decision groups: Supplier Characteristics, On-Going Performance, and Project Management Capabilities. The supplier alternatives are classified by the decision makers corresponding to company size, quality system implementation, and cost management. The application of the proposed approach shows a great accuracy in the final utility values and a significant reduction in the calculation requirements.

本文针对基于传统分析网络过程（ANP）方法的群体决策问题，提出了一种称为“神经模糊分析网络过程（NFANP）”的新型软计算方法。所提出的方法使用移动的，不固定的，梯形和三角形隶属度函数处理模糊环境中判断的区间值，以及由alpha割定义的区间数值比和决策者的置信度。通过允许一定的公差偏差小于0.20，解决了模糊倒数矩阵的一致性问题。此外，建议的方法中包括经过训练的人工神经网络（ANN），以减少将决策因素与替代方法相关联所需的大量算术运算计算。在拟议的实施中，选择问题被定义为三个主要决策组：供应商特征，持续绩效和项目管理能力。决策者根据公司规模，质量体系实施和成本管理对供应商的备选方案进行分类。所提出方法的应用表明最终效用值具有很高的准确性，并且大大降低了计算要求。和项目管理能力。决策者根据公司规模，质量体系实施和成本管理对供应商的备选方案进行分类。所提出方法的应用表明最终效用值具有很高的准确性，并且大大降低了计算要求。和项目管理能力。决策者根据公司规模，质量体系实施和成本管理对供应商的备选方案进行分类。所提出方法的应用表明最终效用值具有很高的准确性，并且大大降低了计算要求。