基于粗集约简的Fuzzy ART设备状态监测方法

Monitoring Method of Fuzzy ART Equipment Condition Based on Rough Set Attribute Reduction

针对复杂设备系统的多工况状态监测问题,提出一种基于粗集属性约简的FuzzyART神经网络状态监测方法。该方法利用粗集的信息决策表和决策矩阵对系统的监测参数进行约简提取,降低FuzzyART输入向量的维数。实例测试结果表明,采用粗集约简提取的监测向量与原监测向量具有相同的监测能力,且提高了网络的学习效率。

A new monitoring method based on Fuzzy ART neural network of rough set reduction is presented aiming at the existing problems in state monitoring for a complex system under multiple loading eases. The method uses rough information deeision-making form and decision-making matrix to choose monitoring parameters for the system, thus can effectively reduce the dimension of Fuzzy ART input vectors. The test result proves that the monitoring vector selected by rough reduction has the same monitoring capability with the original monitoring vector; therefore, network efficiency is improved.