基于二等分放大的故障判别神经网络  

Neural Network of Fault Distinguishing Based on Halve Amplification

作　　者：陈少华[1] 马碧燕[1] 雷宇[1] 桂存兵[1] 

机构地区：[1]广东工业大学自动化学院,广东广州510090

出　　处：《电力科学与工程》2006年第2期25-27,共3页Electric Power Science and Engineering

基　　金：广东省科技厅资助项目(040094)

摘　　要：介绍了一种基于二等分取样,并将预处理数据放大的故障判别神经网络。这种理论的实质类似于放大镜的原理,能更精细地区分2种差距很小的故障。仿真结果表明,与普遍的取样及预处理方式相比,该神经网络的输出更理想。An artificial neural network for fault distinguishing by sampling halve is presented, and the pretreatment data are amplified. The essential of this academic is similar with the principium of magnifying glass. It can subtly distinguish two faults with very little difference. It can be indicated by the experiment result that this neural network acts much more ideally comparing with the other networks using common sampling and pretreatment method.

关 键 词：二等分 故障判别 神经网络 

分 类 号：TK32[动力工程及工程热物理—热能工程]