电测与仪表
電測與儀錶
전측여의표
ELECTRICAL MEASUREMENT & INSTRUMENTATION
2015年
11期
112-116
,共5页
L-M算法%N-W方法%BP神经网络%汽轮发电机%故障诊断
L-M算法%N-W方法%BP神經網絡%汽輪髮電機%故障診斷
L-M산법%N-W방법%BP신경망락%기륜발전궤%고장진단
L-M algorithm%N-W method%BP neural network%turbo generator%fault diagnosis
运用Nguyen-Widrow方法初始化Levenberg-Marquardt算法中的变量参数,在此基础上,L-M算法在训练神经网络中的应用,减少了网络训练时权值和阈值的调整次数,加快了算法收敛速度,保证了误差为误差平面的全局最小值,并将使用N-W方法的L-M算法应用于汽轮发电机的故障诊断。经理论和实践证明该方法提高了故障诊断的准确性和可靠性,并为转动机械提供了有效的故障诊断方法。
運用Nguyen-Widrow方法初始化Levenberg-Marquardt算法中的變量參數,在此基礎上,L-M算法在訓練神經網絡中的應用,減少瞭網絡訓練時權值和閾值的調整次數,加快瞭算法收斂速度,保證瞭誤差為誤差平麵的全跼最小值,併將使用N-W方法的L-M算法應用于汽輪髮電機的故障診斷。經理論和實踐證明該方法提高瞭故障診斷的準確性和可靠性,併為轉動機械提供瞭有效的故障診斷方法。
운용Nguyen-Widrow방법초시화Levenberg-Marquardt산법중적변량삼수,재차기출상,L-M산법재훈련신경망락중적응용,감소료망락훈련시권치화역치적조정차수,가쾌료산법수렴속도,보증료오차위오차평면적전국최소치,병장사용N-W방법적L-M산법응용우기륜발전궤적고장진단。경이론화실천증명해방법제고료고장진단적준학성화가고성,병위전동궤계제공료유효적고장진단방법。
Nguyen-Widrow method is used to initialize the variable parameters of the L-M algorithm, and based on which the application of Levenberg-Marquardt algorithm in neural network training can reduce the number of the ad-justment of the weight and threshold value , speed up the convergence speed of BP algorithm to ensure error to be the global minimum of the error plane , and apply the L-M algorithm combined with N-W method for fault diagnosis of steam turbine .The theory and practice have proved that:this method has improved the accuracy and reliability of fault diagnosis and provided effective method for the rotation machinary fault diagnosis .