CAJ | 학술논문

为了解决智能诊断应用中BP神经网络收敛速度慢、稳定性差以及精度不高的问题，通过嵌入到设备中的诊断Agent采集到设备各元件工作电压，以此为对象研究基于Elman的神经网络故障诊断方法，使用设备故障信息作为BP神经网络和Elman神经网络的训练样本．结果表明，在相同的神经网络训练样本和测试样本下，BP神经网络的收敛速度比Elman神经网络慢，Elman神经网络比BP神经网络诊断精度有提高．经过对训练过程和仿真结果的分析，验证了基于Elman神经网络的故障诊断方法收敛速度提高了约2倍、精确度提高约1．5倍，满足系统在线故障诊断需求．
위료해결지능진단응용중BP신경망락수렴속도만、은정성차이급정도불고적문제，통과감입도설비중적진단Agent채집도설비각원건공작전압，이차위대상연구기우Elman적신경망락고장진단방법，사용설비고장신식작위BP신경망락화Elman신경망락적훈련양본．결과표명，재상동적신경망락훈련양본화측시양본하，BP신경망락적수렴속도비Elman신경망락만，Elman신경망락비BP신경망락진단정도유제고．경과대훈련과정화방진결과적분석，험증료기우Elman신경망락적고장진단방법수렴속도제고료약2배、정학도제고약1．5배，만족계통재선고장진단수구．
Study aims to solve the problems of BP neural network’s low convergence speed ,poor stability and low accuracy in intelligent fault diagnosis .Through the diagnostic Agent embedded into the equipment ,the working voltage across each component is collected .Based on the working voltages ,the fault diagnosis method based on Elman neural network is studied .The equipment fault information is used as the training samples for the BP neural network and Elman neural network .The results show that with the same neural network training samples and testing samples ,the convergence rate of BP neural network is lower than that of the Elman neural network .The diagnostic accuracy of the Elman neural network is better than that of BP neural network .The analysis of the training process and the simulation results shows that the convergence speed of the new method is about 3 times what it was and its accuracy is increased by about 1 .5 times ,meeting the needs of online fault diagnosis systems .