摘要
文中就BP神经网络的学习算法进行了分析,使用PSPICE软件对电路进行仿真,在此基础上建立了模拟电路的BP神经网络故障字典库。通过查阅字典库实现模拟电路的故障诊断。指出了BP神经网络在模拟电路故障诊断过程中存在的网络泛化性能差的问题,并提出了解决办法。经过仿真实验验证,这种方法用与模拟电路故障诊断相结合能够提高神经网络检测故障的智能性,提高故障诊断的准确性,结果较好,具有一定的应用价值。
The algorithm of BP neural network is analyzed and the analog circuit is simulated by using PSPICE software in this paper. Based on these, the fault dictionary is established, and analog circuit fault diagnosis is realized by looking up fault dictionary. The shortage of BP network in diagnosing fault is pointed out and brought up solution. This solution method improves the automation degree to establish diagnostic model and the accuracy of fault diagnosis. The result is better and value by simulation.
出处
《电子测量技术》
2006年第6期205-206,211,共3页
Electronic Measurement Technology
关键词
模拟电路
故障诊断
泛化
BP算法
analog circuit
fault diagnosis
generalization
BP algorithm
