基于小波分析与人工神经网络的水轮机压力脉动信号分析  

Analysis of pressure fluctuation in draft tube based on wavelet analysis and artificial neural networks

作　　者：赵林明[1] 楚清河[2] 代秋平[1] 王利英[1] 

机构地区：[1]河北工程大学,河北邯郸056021 [2]华北水利水电学院,河南郑州450011

出　　处：《水利学报》2011年第9期1075-1080,共6页Journal of Hydraulic Engineering

基　　金：国家自然科学基金项目(60940036);河北省自然科学基金项目(E2010001026)

摘　　要：针对水轮机尾水管压力脉动信号的非平稳和时变特性,提出了一种基于小波分析和自组织人工神经网络相结合的尾水管压力脉动信号的分析方法。这种方法首先应用小波阈值法对信号进行降噪减少干扰,然后将小波分解系数重构得到不同频带的信号分量,并提取显著的不同频带能量,最后将各频带能量作为特征向量,用自组织人工神经网络进行模式识别,得到了尾水管压力脉动的不同模式。应用该方法对某混流水轮机的压力脉动试验结果进行了分析,结果表明,该分析方法是有效的,能够对水轮机尾水管中的压力脉动状态进行有效的识别。In view of the non-stationary and time-varying characteristics of the pressure fluctuation signal in draft tube,this paper presents a method combining wavelet analysis with a self-organizing artificial neural network to analysis the pressure fluctuation signal.Firstly,the wavelet threshold value method was used to decrease the noise and reduce interference,then the wavelet coefficients were reconstructed to obtain signal component of different frequency band and extract significant different band energy.Then,the band energy is used as the characteristic vector to apply the self-organizing neural network for pattern recognition and obtained the different patterns of pressure fluctuation in draft tube.This method was used to analyze the pressure fluctuation data for a model of Francis turbine.The results show that this method is effective in identifying the state of pressure fluctuathon in draft tube.

关 键 词：水轮机 小波分析 自组织人工神经网络 模式识别 

分 类 号：TV131.63[水利工程—水力学及河流动力学]