基于人工免疫系统的电路小样本故障诊断方法

针对常规电路故障诊断方法存在的故障样本需求大的问题,基于免疫系统的阴性选择机理,提出了建立人工免疫系统实现电路小样本故障诊断的新方法。该方法仅需电路正常模式测试数据加一组故障模式测试数据为样本,生成随机检测器,再运用变异实值否定选择算法优化随机检测器,结合基于故障样本生成故障类型检测器构成人工免疫系统的故障检测器,并采用二次匹配方法完成电路的故障诊断。对ITC’97的CTSV滤波器电路的故障诊断表明,该方法可应用于小故障样本场合下的电路故障诊断,具有较高的实际应用前景。

Fault diagnosis of circuit in small sample based on artificial immune system

In order to solve the problems such as large sample demands in the fault diagnosis techniques of circuit,a new method based on immunology principles is proposed.Only one set of fault-free circuit and fault circuit＇s data are selected as the input of the artificial immune system（NS） to generate the random detector set.And the mutation based real-value negative selection algorithm is applied to the optimization of the volume covered by the random detector set.Therefor,the optimized random detector set and the remember detector set directly generated by known fault states constitute the antibody of the artificial immune system—the fault detector set.Finally,the fault diagnosis method based on two-time-matching NS is applied to diagnose the fault of circuit.The proposed method was applied to the fault diagnosis of Continuous-Time State-Variable Filter（ITC＇97 benchmark circuits）.The experiment results show the artificial immune system based on random real-value negative selection algorithm has good application prospects among the fault diagnosis of circuit in small sample.