CAJ | 학술논문

为了实现某型号坦克传动系统齿轮箱的智能化故障诊断，采用带偏差单元的递归神经网络理论，建立了基于带有偏单元的递归神经网络的齿轮箱故障诊断模型。针对某型号坦克传动系统齿轮箱的主要故障，将数据采集器现场采集到的齿轮箱数据作预处理后的数据和人为设置了一些故障的数据分别用带偏差单元的递归神经网络模型进行了分析和判断。结果表明，带偏差单元的递归神经网络在本型号坦克齿轮箱的故障诊断中能取得满意的效果，算法精度高，收敛性好。
위료실현모형호탄극전동계통치륜상적지능화고장진단，채용대편차단원적체귀신경망락이론，건립료기우대유편단원적체귀신경망락적치륜상고장진단모형。침대모형호탄극전동계통치륜상적주요고장，장수거채집기현장채집도적치륜상수거작예처리후적수거화인위설치료일사고장적수거분별용대편차단원적체귀신경망락모형진행료분석화판단。결과표명，대편차단원적체귀신경망락재본형호탄극치륜상적고장진단중능취득만의적효과，산법정도고，수렴성호。
In order to realize the intelligentization of fault diagnosis of gearbox in a certain type of tank drive system,the gearbox fault diagnosis model based on a recurrent neural network with bias unit was established according to the theory of recur-rent neural network with bias unit. Aiming at the main faults in the gear box in a certain type of tank drive system,the data that collected by collector was preprocessed and the data of some faults set by people were analyzed by using recurrent neural net-work model with deviation unit. The results show that the recurrent neural network with bias unit can achieve satisfactory effect in the fault diagnosis for gear box in this type of tank drive system. The algorithm has high accuracy and good convergence. The innovation point is that the gearbox fault diagnosis model based on a recurrent neural network with bias unit has been used for the first time in the gearbox fault’s intelligent diagnosis in a simulation experiment.