基于AHP与重要性指标筛选的神经网络评价模型与应用

以对评价目标有影响的所有评价指标作为神经网络的输入,会导致网络模型复杂、降低其性能和影响计算精度的问题,因而提出基于层次分析法和重要性指标筛选法的神经网络评价建模方法即首先运用层次分析法对评价指标进行重要度排序,然后利用重要性指标筛选法过滤出对评价目标有重要影响的指标,以其结果作为神经网络的输入。该法不仅简化网络模型,而且提高网络的性能和计算精度。运用该法对企业安全工作评价,结果证明,不仅是可行的,而且达到了预期的目的。

ANN Assessment Model and Its Application Based on AHP and the Sieving of Important Index

Regarding the entire indices affecting assessing objective as the input variables of ANN (Artificial Neural Network) model not only makes the model very complex, but also greatly lowers the capacity and calculation accuracy of this model. To solve this problem, a new method for establishing ANN model is presented based on AHP (Analytic Hierarchy Process) and the sieving of important index. The establishment steps are as follows: firstly the assessing indices are put in order according to their importance in assessing objective by using AHP, secondly all the important indices greatly affecting the objective are selected by the method for sieving important index, then the indices obtained are regarded as the input of ANN, thus a new model is set up, which greatly improves the capacity and calculating accuracy of ANN. Finally, the application of this model in the assessment of enterprise' safety work is given; the result shows that this method is feasible and overcomes the shortages described above.