基于RNN模型的高科技企业经营困境预测  

Prediction of High-tech Enterprise′s Financial Distress with the Multi-factors Based on RNN Model

作　　者：尹鹏[1] 王宗军[1] 肖德云[2] 薄纯林[3] 

机构地区：[1]华中科技大学管理学院,湖北武汉430074 [2]武汉理工大学经济学院,湖北武汉430070 [3]交通银行审计部,上海200120

出　　处：《武汉理工大学学报（信息与管理工程版）》2010年第5期815-819,共5页Journal of Wuhan University of Technology：Information & Management Engineering

基　　金：国家自然科学基金资助项目(70872033)

摘　　要：指出了现有企业经营困境预测模型和指标体系的不足;针对高科技企业的特性,通过结合智力资本、外部环境因素及传统财务指标构建相对完备的预测指标体系,将基于粗糙集和神经网络相耦合的粗集神经网络模型作为预测方法,对92家高科技企业组成的检验样本进行实证分析,结果表明,该模型的预测正确率为96.7%,比神经网络的预测精度更高,且所需训练时间更短,生成规则更少,具有较大的实用价值。Disadvantages of some prediction models and indicator system of enterprise financial distress were figured out;and a more completed prediction indicator system was constructed by adding the intellectual capital and outer external environment indicators.The RNN（rough neural network） prediction model is a hybridism of rough set theory and neural networks.When the model was applied to predict a sample which included 92 high-tech enterprises,the accuracy of prediction was 96.7%.The RNN model also shortened the training time and generated fewer rules than the NN model.Therefore,it could be considered as a realistic prediction tool.

关 键 词：经营困境预测 智力资本 外部环境 粗集神经网络 

分 类 号：F272.1[经济管理—企业管理]