Kohonen网络在汽轮机振动故障诊断中的应用

The Application of Kohonen Network to the Diagnosis of Oscillation Breakdowns in Steam Turbine

利用Kohonen网络聚类的特点,把汽轮机振动故障信号频谱中的相关频段上不同频率谱的谱峰能量值作为故障信号的训练样本输入到Kohonen网络,并由网络进行聚类,产生聚类中心点。根据此聚类中心点的位置来确认和诊断汽轮机振动故障的原因以及目前的严重程度。

With the feature of clustering in Kohonen Network ,input the energy amplitudes of various of harmonic spectrums about signals of oscillation breakdowns in steam triblet to Kohonen Network as the training sample of breakdowns signals, and cluster by Kohonen Network, generating the clustering central points .According to the position of such clustering central points ,the reason for vibration breakdowns in steam turbine and the degree of gravity can be affirmed and diagnosed.