水电站厂房结构振动主要是水力、机械和电磁三大类振源引起的,厂房结构与机组之间存在明显的耦联作用,厂房结构和机组振动系统呈明显的动态耦合效应和非线性特征。本文在对实测厂房结构与机组振动响应关系分析的基础上,提出应用神经网络预测方法可在不考虑结构精确的数学力学模型的前提条件下,得出被研究对象的非线性振动特性,即通过机组的振动和尾水脉动的监测数据预测厂房结构的振动。该方法应用于三峡水电站厂房结构的振动预测,结果与实测结果基本吻合。
The vibration of powerhouse structures may be induced by hydraulics factors, mechanical and electromagnetic factors of the generating unit. It nonlinearly couples with the generating unit. On the basis of part of observation data, the artificial neural network method is applied to predict the vibration of power house in the Three Gorges Project according to the vibration of generating unit and water pressure fluctuation in draft tube. The validity of the predicted result is verified by another part of observation data.