基于指标规范值的室内空气客观评价的概率神经网络模型

为了建立简单、普适、通用的概率神经网络的室内空气评价模型,在适当设定室内空气各项指标的参照值及指标值的规范变换式基础上,使室内空气同级标准不同指标的规范值差异尽可能小,从而用规范值表示的各指标都可用同一个规范指标"等效"替代。因此,概率神经网络隐层各类模式的基函数中心矢量的各指标分量值与同级标准所有15项指标规范值的均值等同。将基于指标规范值的概率神经网络模型用于室内空气的评价实例进行检验,验证了该模型的普适性、通用性和简便性。

Model of objective indoor air quality evaluation with normalized indices values based on probabilistic neural network

In order to establish the indoor air quality evaluation model based on probabilistic neural network, which is of simple, universal and general, with properly set reference values and transformed forms for each indoor air index, the difference in the same grade standard values with different air indexes could be as small as possible after the normal transformation, so the normalized values of different indexes were equivalent to a same normalized index. Therefore, the each index component values of the center vector of basis function for the hidden layer modes of probabilistic neural network equal to the mean of normalized values for all 15 item indexes of the same grade standard. The indoor air quality evaluation model based on probabilistic neural network with normalized indexes values(NV-PNN) is used for the evaluations of indoor air quality cases and the results show that the NV-PNN model has the characteristics of universality, generality and simplicity.