Abstract: Aiming at the complex relationship among the elements of customer requirement in software development, the defects of AHP and FAHP in solving the complex relationship of the demand are analyzed. This paper uses analytic network process and trapezoidal fuzzy number, establishes the ultimate weighted super matrix to get the mixed weights of elements of the demand. The weights of fuzzy analytic network process, fuzzy analytic hierarchy process and analytic network process are compared and analyzed. It is expounded that the interrelation of elements in complex systems and the fuzzy evaluation of relative importance of the customer demand play an important role in the process of solving the weight, and the feasibility of fuzzy network analysis in solving complex demand weight is verified. The results provide a basis for the accurate solution of the complex weights of customer requirements in software development.

Key words: demand weight, analytic network process, trapezoidal fuzzy number, fuzzy analytical hierarchy process, fuzzy analytic network process

摘要： 针对软件开发中顾客需求元素间的复杂关系，分析了层次分析法（AHP）与模糊层次分析法（FAHP）在解决复杂需求权重时所存在的缺陷；运用网络分析法（ANP）与梯形模糊数相结合的模糊网络分析法，建立加权极限超矩阵求解需求元素的混合权重；对比分析了模糊网络分析法、模糊层次分析法和网络分析法的权重结果；揭示了在复杂系统中元素的相互联系和顾客需求相对重要度评价的模糊性在权重的求解中的重要性；验证了模糊网络分析法在复杂需求权重求解中的可行性。研究结果为软件开发中顾客需求复杂权重的准确求解提供了依据。

关键词: 需求权重, 网络分析法, 梯形模糊数, 模糊层次分析法, 模糊网络分析法