作　　者： ; ;

机构地区： 中山大学软件学院

出　　处： 《计算机仿真》 2012年第1期352-355,共4页

摘　　要： 由于纳税评估过程中存在不精确、模糊以及冗余信息,传统评估模型多数采用经验法和比较法,缺乏科学性和公正性,评估结果正确率低。为了提高纳税信用等级评估的正确率,提出了一种采用模糊神经网络的纳税信用等级评估模型。首先利用模糊逻辑推理对纳税评估过程中的不精确、模糊的信息进行有效的处理,然后利用训练数据对神经网络模型进行训练学习,获得纳税评估指标和信用等级间的评估模型,最后通过利用测试集对模型进行验证,结果表明,模糊神经网络方法提高了纳税信用等级评估的正确率,为税收信用评估提供有效的依据。 Research tax payment evaluation grade of enterprise. The traditional evaluation models are inefficient, and the results accuracy is not high. In order to get better tax payment evaluation grade, this paper presented a fuzzy neural network tax enterprise credit evaluation grade model. Firstly, fuzzy logic reasoning was used to deal with information whieh is not precise and fuzzy. Scondly, training data were used to train the fuzzy neural network model, to get the relation knowledge of the enterprise evaluation index and credit rating classification. Finally, the experiment was designed to verify the model. The results show that the fuzzy neural network has good prediction performance of tax assessment.

关 键 词： 模糊理论 神经网络 纳税评估

领　　域： [自动化与计算机技术] [自动化与计算机技术]