基于蚁群神经网络的财务危机预警方法

Ant Colony Algorithms and Neural Network Based Early-warning System for Enterprise Financial Distress

为了克服神经网络财务危机预警方法收敛慢、不收敛和网络结构难以确定等缺陷,提出了基于蚁群算法的改进神经网络财务危机预警方法。将神经网络模型的结构和参数进行编码,利用蚁群算法确定若干个神经网络模型的结构和参数,然后通过评价函数得到神经网络的最佳结构,最后通过BP算法训练该神经网络,得到神经网络财务危机预警模型。验证结果表明,该模型结构简单、预警精度高。

In order to overcome the defects of the neural network early-warning system for enterprise financial distress including slow convergence,non-convergence and difficulty in determining network structure, the improved neural network early-warning method for enterprise financial distress based on ant colony algorithm is presented.First,the structure and parameters of the neural network is encoded. Some structures and initiating parameters of neural network are obtained by the ant colony algorithm. Then,the optical structure of neural network is determined by means of the evaluation function.Finally, the neural network is trained through BP algorithm and the neural network early-warning model for enterprise financial distress is got.The experiment results show that the model has simple structure and lower error rate.