基于多属性决策网MADN的仿真系统VV&A理论方法研究

仿真系统能够正确应用的前提是拥有足够的可信度，VV&A(Verification, Validation & Accreditation; 校核、验证与确认)是评估仿真可信度的重要技术。为了弥补传统VV&A方法的不足，本课题提出一种新的多属性决策网MADN（Multiple Attribute Decision Network）理论，建立和研究网络的数学基础、规则、图示、各分析和计算方法等，并基于MADN开展仿真系统VV&A的各关键技术研究，包括：研究网型的多维评估指标体系以替代传统的树型评估指标体系、研究非线性指标综合算法以替代线性指标综合算法、研究VV&A分析结果至仿真可信度的转换方法以完成评估指标在标度空间内的统一、研究系统可信度缺陷环节的检测方法以实现VV&A的可回溯性。目的是更科学地计算仿真系统的可信度，以支持仿真的正确应用和自身的修正。

It is a prerequisite for simulation systems to have adequate credibilities before their using, and VV&A (Verification, Validation & Accreditation) is the key technique for simulation credibility evaluation. In order to make a breakthrough for conventioanal VV&A evaluation methods, a new theory of MADN (Multiple Attribute Decision Network) is proposed, and the mathematics foundation, rules, diagrams, analyzing and calculation methods etc. of the network are established and researched. Furthermore, based on the MADN theory some key techniques of VV&A are researched, which include net-shaped multiple dimension evaluation index systems substituting for index trees, non-linear composite index arithmetic substituting for linear arithmetic, transition methods for VV&A analysis results to simulation credibility to unify evaluation indexes in a scale space, and searching method for system's credibility fault parts to realize the traceability of VV&A. The project aims to compute the credibility of simulation systems more scientifically, and support the right application and self-modification of the simulations.