模拟电路故障检测与定位新方法

A New Method for Fault Detection and Location of Analog Circuits

提出了一种基于统计理论与智能信息处理技术的容差模拟电路故障检测与定位新方法。在充分考虑容差效应的基础上,构建了故障阈值函数与故障判据,从而可通过监测可测点工作电压实现电路的故障在线检测。再通过离线测量电路在不同测试频率下输出对输入的增益,将可测点工作电压与电路增益两类测试信息经特征层融合,运用所提出的遗传神经网络方法对电路进行故障定位。仿真结果表明:所提方法对模拟电路的硬故障与元件参数偏移较小的软故障均适用,故障检测与定位准确率高。

Based on statistics theory and intelligent information processing technology, a new method for fault detection and location in analog circuits with tolerance was presented. Tolerance effort to terminal performance was calculated according to element parameter deviation, by which a fault threshold function and a fault criterion were established. By monitoring accessible node voltages, fault on-line detection was implemented based on the proposed fault threshold function and fault criterion. Then, circuit gains under different test frequencies were gained by off-line measurement. Under the full consideration of fault and tolerance effort, fault feature was extracted from node voltages and circuit gains and characteristic data fusion was realized. Fault location was performed by means of the proposed GA-BP algorithm. The simulation results show that the proposed method can detect and locate catastrophic and parametric faults of tolerance circuits and has satisfactory diagnosis accuracy.