RS_RBF在高新技术项目投资风险评价的应用

Study on Risk Evaluation of High-tech Projects Investment Based on RS_RBF

研究高新技术项目投资风险评价问题。由于高新技术项目各评价指标间的存在着冗余信息,传统的方法无法除去这些冗余,导致风险评价结果有时不正确,出现误导信息。为了提高风险评价的正确性,提出了一种粗糙集(RS)和径向基函数(RBF)神经网络相结合的高新技术项目投资风险评价混合模型。首先利用粗糙集的强大的数值分析能力对各评价指标进行属性约简,从而减少了RBF神经网络的训练数据,简化了网络结构,然后对约简后的数据利用RBF神经网络进行训练,最后应用于高新技术项目投资风险评价中。仿真结果表明,与RBF神经网络模型相比,混合模型加快了网络的运算速度,评价误差更小,评价精度进一步提高,获得了较好的评价结果。

In order to resolve the redundant information in High-tech project evaluation,rough sets based RBF neural network nodel are presented for the evaluation model of high-tech investment risk..With the powerful numerical analysis capabilities of rough set,the attributes of evaluation indexes are reduced,which decreases the training data of RBF neural network,thereby simplifies the structure of RBF neural network and speeds up the training speed.Simulation results show that the hybrid model can achieve more satisfactory results,and compaired with the standard RBF neural network model,the hybrid model has obvious advantages.