大规模模拟电路故障定位的快速算法及实现

Algorithm for Fast Fault Location of Large--Scale Analog Circuits and Its Realization

要使现有的模拟电路故障诊断算法适用于大规模电网络,必须将大规模网络分解成互不耦合的小规模网络。文献[1]给出了一种快速定位算法(算法Ⅰ),但未考虑容差。本文从解决模拟电路故障诊断中最棘手的问题——容差、在线计算量和离线计算量的目的出发,将故障预测技术、替代理论和迭加定理运用到支路撕裂法中,从而得到一个新的大规模模拟电路的快速故障定位算法(算法Ⅱ),同时对算法Ⅰ、Ⅱ进行了FORTRAN编程实现。研究结果证明,算法Ⅰ、Ⅱ适用于小、大规模网络的单、多故障诊断。当网络被连续监测时,算法Ⅱ的稳健性、计算量和存贮器要求均较算法Ⅰ有较大改进。文中举了一个有源电路的实例,验证了上述论断的有效性。

To make the existing fault diagnosis algorithm suitable and feasible for practicallarge analog networks, a large network must be decomposed into small uncoupled subnetworks.The fast aleorithm (algorithm I), which is an approach not considering robustness for fault diag-nosis of large analoge circuit by byarich teering, was given in [1]. Here, in order to solve thesequestions, namely the uobustness and the computation time, which are very difficult in analogfaults diagnosis, a new fast algorithm (algorilhm Ⅲ) for fault location of large analog circuits us-ing fault prediction technique, substitution, superposition theorem and branch teering method issuggested. The algorithm Ⅰ and Ⅲ are programmed using FORTRAN and realized on microcom-puter. The results of this study show that those algorithms can be applied to small or large analogcircuits for the location of smigle and mutli--faults. When the network can be continuosly moni-tored, algorithm Ⅱ resutls a significant improvenent in both robrstness, computation time andmemory requirement over the algorithm Ⅰ. Demonstrative example of an active circuit is given toshow the effectiveness of the proposed method.