There is nonlinear relation between the commercial bank credit risk grade and the evaluation index, and it is difficult using the traditional models to evaluate the credit risk's grade. The BP neural network has good non - linear approximation ability, but the BP neural network only chooses the quantitative indices, so the evaluation in-dex is not comprehensive. In order to improve the commercial bank credit risk assessment, the model of AHP and the BP neural network were established. Firstly, the analytical hierarchy process was used to obtain the weights of each indexes, then according to the size of the index weight sorting, the index with smaller weight was eliminated to sim-plify the BP neural network structure. Then the BP neural network was trained and verified. Experimental results show that compared with the traditional BP neural network model, the combined model can simplify the model,, im-prove the operation efficiency and speed, and enhance appraising precision.%研究商业银行信用风险准确评估问题,由于商业银行信贷资金安全存在不确定性,信用风险评估指标较多,指标间存在大量重复信息,风险等级与指标间呈非线性关系,导致传统评估模型很难精确地进行评估,评估精度不高.为了提高商业银行信用风险评估精度,提出一种将层次分析法(AHP)与BPNN(BPNN)相结合的的商业银行信用风险评估模型(AHPBPNN).模型首先利用层次分析法求出各指标的权重,并按照权重的大小进行指标排序,消除指标的重复信息,使评估指标得到了精简,然后将经过处理后的指标输入BPNN,通过进行非线性学习和建模,最后对信用风险进行评估仿真.实验结果表明,AHP-BPNN简化了评估指标体系,提高了评估的速度和精度,增加了商业银行信用风险评估的有效性.
展开▼
