基于改进液体状态机的模拟电路故障诊断  

Analog Circuit Fault Diagnosis Based on the Improved Liquid State Machine

作　　者：周世官[1] 李国君[1] 郑燕[1] 

机构地区：[1]宁波工程学院电子与信息工程学院,浙江宁波315016

出　　处：《电子科技》2017年第5期5-7,11,共4页Electronic Science and Technology

摘　　要：针对标准液体状态机存在的过层次化和病态同步两个基本缺陷,结合遗传算法和液体状态机的特点,提出了一种基于改进液体状态机的故障诊断方法。该方法首先在液体状态机训练前运用遗传算法对标准液体状态机液体池中的大规模随机稀疏网络进行迭代优化,然后采用改进液体状态机进行模拟电路故障模式的分类。仿真表明,和标准液体状态机相比,改进液体状态机可以有效地提高模拟电路故障诊断的准确率,对模拟电路的故障诊断研究有一定的实际意义。A fault diagnosis method based on the improved liquid state machine is proposed to solve the problem of pathological synchrony and over - stratification in the standard liquid state machine, which is based on the characteristics of genetic algorithm and liquid state machine. The method firstly uses the genetic algorithm to iteratively optimize the large - scale stochastic sparse network in the standard liquid state machine liquid pool before the liquid state machine training, and then uses the improved liquid state machine to classify the fault pattern of analog circuits. The simulation results show that compared with the standard liquid state machine, the improved liquid state machine can effectively improve the accuracy of analog circuit fault diagnosis.

关 键 词：液体状态机 遗传算法 故障诊断 模拟电路 

分 类 号：TN431.1[电子电信—微电子学与固体电子学] TP18[自动化与计算机技术—控制理论与控制工程]