基于粗糙集与模糊神经网络的故障诊断

神经网络已广泛应用于故障诊断领域,但也存在冗余数据难以剔除、网络结构复杂的缺陷,与其他理论相结合可以得到更优越的诊断特性。粗糙集在数据约简和规则获取等方面具有优势,可以有效避免神经网络构造的困难;同时,模糊神经网络可以使网络推理过程变得透明,且较强的数据泛化能力可以弥补粗糙集的不足。通过将粗糙集与模糊神经网络技术相结合,建立故障诊断模型,利用歼击机操纵面故障诊断实例检验其可行性和有效性。

Fault Diagnosis Based on Rough Set and Fuzzy Neural Network

Neural network has extensively applied to fault diagnosis. By being combined with other technologies, it will achieve more superiority over its defect in redundancy of data and network construct. Rough set theory has the advantage of data reduction and rule extraction, and can initialize the structure of network. The tolerances to data of fuzzy neural network can improve the application of rough set theory; and it also provides a clear reasoning. By combining those advantages of the two technologies, the model of fault diagnosis was constructed, and the results of the simulation for fighter plane control surface fault diagnosis revealed the efficiency and practicality of this method.