广东电力2018，Vol.31Issue(5)：1-6,6.DOI:10.3969/j.issn.1007-290X.2018.005.001
基于连续小波变换和卷积神经网络的尾水管涡带状态识别
State Identification of Draft Tube Vortex Based on Continuous Wavelet Transform and Convolutional Neural Network
摘要
Abstract
Pressure pulsation in the draft tube of the Francis hydroturbine is the important factor affecting stability of the hy-droelectric generating set. It is necessary to monitor and identify state of draft tube vortex for ensuring safe and stable oper-ation of the hydroelectric generating set. Therefore,this paper uses wavelet coefficient cloud chart which can effectively re-present time-frequency domain characteristics of signals as the feature image,combines good adaptability of the convolution-al neural network (CNN)to the topology of image and introduces intelligent image identification technology into state iden-tification for the draft tube vortex. Furthermore,it presents a kind of state identification method for the draft tube vortex based on continuous wavelet transform and CNN which has realized automatic extraction of textural features of time-fre-quency domain charts,avoided artificial identification and simplified preprocessing program. This method has rapidly and accurately identified state of the draft tube vortex. According to experimental data of variable load of one 200 MW Francis hydroturbine in a hydropower station,it verifies effectiveness of the proposed method.关键词
尾水管涡带/状态识别/小波系数云图/连续小波变换/卷积神经网络Key words
draft tube vortex/state identification/wavelet coefficient cloud chart/continuous wavelet transform/convolu-tional neural network分类
能源科技引用本文复制引用
洪礼聪,王卫玉,陈启卷..基于连续小波变换和卷积神经网络的尾水管涡带状态识别[J].广东电力,2018,31(5):1-6,6.基金项目
国家自然科学基金项目(51379160) （51379160）
