摘要
针对模拟电路中部分故障类别发生重叠的特点,提出了一种基于量子神经网络算法的模拟电路故障诊断方法。在被测电路输出端采集时域响应信号,计算其峭度和熵,作为特征量,并应用量子神经网络算法对模拟电路的各个不同的故障类别进行辨别。实验结果表明,构建的神经网络具有简单的网络结构,且故障诊断正确率较高,达到99.62%。
To solve the overlap of part of fault classes in the analog circuit fault diagnostics,a novel analog circuit fault diagnostics approach based on quantum neural networks algorithm was presented.Kurtosis and entropy were calculated as features after the time domain response signals of the circuit under test were measured,and then the different fault classes were identified by quantum neural networks algorithm.The simulation demonstrated that constructed neural network had simple network structure and the fault diagnosis accuracy was higher,which reached 99.62%.
出处
《辽宁石油化工大学学报》
CAS
2015年第2期58-61,共4页
Journal of Liaoning Petrochemical University
基金
国家杰出青年科学基金项目(50925727)
国防科技计划项目(C1120110004
9140A27020211DZ5102)
国家自然科学基金项目(61102035
61401139
61403115)
教育部科学技术研究重大项目(313018)
安徽省科技计划重点项目(1301022036)
关键词
模拟电路
故障诊断
峭度
熵
量子神经网络
Analog circuit
Fault diagnostics
Kurtosis
Entropy
Quantum neural networks