基于变精度粗糙集的水电机组故障诊断

将变精度粗糙集理论引入水电机组故障诊断中,利用变精度粗糙集属性约简方法对水电机组故障的检测信息进行约简,提取对故障分类起主要作用的信息,并用RBF神经网络对粗糙集处理后的故障信息进行诊断。该方法不仅克服了神经网络对冗余信息和有用信息识别的局限性,有效地降低了神经网络的输入信息空间维数,减小了神经网络规模,还可以弥补经典粗糙集方法对输入信息中的噪声较敏感、抗干扰能力差的不足,进而达到提高诊断准确性的目的。水电机组振动故障实例的诊断分析结果证明了该诊断方法的有效性和优越性。

Research on the fault diagnosis of hydroelectric units based on variable precision rough set

The variable precision rough set theory is introduced into the fault diagnosis of hydroelectric units.The inspection information of hydroelectric units' faults is reduced by the reduction method of the variable precision rough set,and the information that plays a major role in fault classification can be obtained.The information treated by the rough set is diagnosed with a radial basis function(RBF) neural network.The method not only bypasses the limitations of neural networks,which may fail to determine the redundancy or usefulness of information,it also decreases the spatial dimension of the network input information effectively,reducing the scale of the neural network.The method makes up for the defects of the classical rough set,such as the sensitivity to noise of input information and the weak anti-jamming capability,and accordingly improves the accuracy of fault diagnosis.The result of a fault example proves that the proposed method is valid and has significant advantages.