基于独特型免疫网络的故障诊断方法研究

随着目前大型设备的日益复杂，传感器节点之间的数据是动态互相影响的，如果仅靠从知识库中去寻找最佳匹配来诊断故障，很容易发生漏诊或误诊。根据Jerne的独特型网络理论提出了一个简单、通用性好，能综合反映网络细胞（抗体和抗原）的浓度、相似度信息，以提高免疫算法的全局搜索能力的模型，该模型利用案例检索的优势，考虑了机电设备结点之间相互影响的特性。实验证明，该模型在加快免疫算法后期的收敛速度和提高故障检测率方面均高于其他算法。

Study on Fault Diagnosis Methods based on Idiotypic Immune Network

Existing large equipments become increasingly complicated,data among transducer nodes are influenced each other dynamically.Diagnosis error or misdiagnosis will occur if fault diagnosis is only dependent on optimal matching from the knowledge library.This paper give us a simple model which can synthetically reflect the density of network cell(antibody and antigen) and the information of alike-degree between them according to Jerne's Idiotypic Network theory,interactive characteristics between transducer nodes are considered in the model and advantage of CBR is utilized.Experiments show that this model is better than other immune algorithms in speeding convergence ratio of later period of the immune algorithms and improving detection rate of the fault diagnosis.