【摘要】 研究了基于多准则学习的模糊神经网络评价环境大气质量的模型。将环境质量评价标准作为模糊集 ,采用多输出神经网络 ,得到一个实际输出向量 ,把它作为评价样本对该模糊集的隶属度 ,较好地克服了常用的单输出网络人为规定评价指数的主观因素 ;把隶属度向量作为权值 ,对评价样本进行综合评分 ,以此对环境质量进行评价 ,避免了当隶属度向量各分量分布不集中时 ,最大隶属度原则所遇到的困难。采用基于多准则学习神经网络 ,较好地克服了基于单准则学习神经网络收敛速度慢 ,易陷入局部极小值的缺点更多还原

【Abstract】 This paper is concerned with an assessment model of air environmental quality by using fuzzy neural network with multi criteria learning algorithm.Regarding the evaluation criterions of air environmental quality as fuzzy sets and using multi output neural network to acquire a actual output vector and taking it as degree of membership of evaluation sample to the fuzzy set,the subjective affect in establishing evaluation index of traditional single output network can be overcome.When take the degree of membership vector as weight value and grade the evaluation sample synthetically to evaluate air environmental quality,we can avoid the difficulty in maximum degree of membership criterion when the component of degree of membership vector is not concetrated.The disadvantage of converging slowly and liable to local minimum when adopting single criterion learning neural network could get over by using multi criterion learning neural network.更多还原