摘要
SOM神经网络是一个由全连接的神经元阵列组成的无教师自组织、自学习网络,它具有很强的非线性映射能力和柔性网络结构以及高度的容错性和鲁棒性。通过反复多次迭代训练,使之对训练样本形成科学合理的分类,然后利用此网络结构对待测样本进行实例仿真检验,判断此设备是否处于异常状态,从而可对设备的运行状态进行实时监测与智能诊断。实际算例也表明了该理论在内燃机智能故障诊断中的有效性及实用性。
SOM neural network is full-connection nerve cell array for self-organizing and self-learning without teachers. It is of strong non-linearity mapping capacity and flexible network structure. Through training and training again, SOMNN sets up scientific and rational classification to trained examples; then the trained network structure is used to simulate those waiting tested examples ; at last, judge the device is of abnormal status or not. The instance indicates the method is of validity and practicability.
出处
《小型内燃机与摩托车》
CAS
北大核心
2009年第4期49-51,共3页
Small Internal Combustion Engine and Motorcycle
关键词
SOM神经网络
内燃机故障
智能诊断
研究
SOM neural network, I. C. engine malfunction, Intelligence diagnosis, Research
