基于Hamming神经网络的通信网可靠性指标评价模型

Evaluation Model of Reliability Index on Communication Network based on Hamming Neural Networks

1.引言目前,多目标综合评价方法中计算各评价指标权重的方法根据计算权系数时原始数据的来源不同可以分为两大类:一类是主观赋权法,其原始数据主要根据评价者主观上对各指标的重视程度来决定权重的方法,如吉林法、AHP法、专家调查法等。目前,对于主观赋权法的研究比较成熟,该类方法的优点是专家可以根据实际问题,较为合理地确定各指标之间的排序,也就是说尽管主观赋权法不能准确确定各指标的权重,但在通常情况下可以在一定程度上有效地确定各指标按重要程度给定的权重的大小顺序,同时解释性好。它的主要缺点是主观随意性大,而且这一点并未因采取诸如增加专家数量、仔细挑选专家等措施而得到根本的改善,因而。

On synthetic evaluation of multiobjective,how to calculate indexs' weight is a key problem. As a neural network has neuro-computing. parallel distributed processing and self-adaptive learning characters, and can implement non-linear mapping, making use of neural networks to evaluate indexes system of reliability on communication networks is very effective and convenient. Using Hamming neural network to evaluate reliability indexes of communication networks,it is convenient that for a complicated topology networks to evaluate reliability by a practical application example.