中文摘要：

为克服商业银行信用风险评价中所遇到的模糊综合评判失效及警限确定的难题，通过能量极小点的设计，利用H.pfield神经网络记忆与联想功能，建立基于H.pfield神经网络的风险评价模型。将其应用于信用风险评价，网络运行结果可以反映信用风险的当前状态。研究还表明，该模型能在一定程度上反映样本数据的数字特征，适合于信用风险的评价，但其评价能力受记忆容量及样本差异的影响。

英文摘要：

In order to avoiding the invalidation of the fuzzy synthetic judgment and overcoming the difficulty to make early-warning line of the existing models to evaluate the credit risk, the evaluating model based on Hopfield neural network is found by using the association function of network through the design of the weight to set the given models in the network. When the model was applied to the evaluation of the credit risk, the risk degree can be obtained. At the same time, the characters of the original data can be reflected by it. Study testifies to the feasibility and practicability of the evaluation model. However, the evaluation effect of the model is under control of the memory capacity and the difference of the sample.