摘要
针对电梯实时故障诊断困难及故障点定位准确率低等缺点,提出一种基于神经网络且结合了D-S理论的信息融合诊断方法,并建立了相应的故障诊断模型。对小波包特征参数提取的原理、过程及特点作了详尽的描述与分析,将其应用到电梯轿厢振动信号的分解与提取,并将所提取出来的内容作为特征向量,再结合神经网络与D-S证据理论,进行故障诊断。进一步介绍了BP神经网络训练的过程,结合实验仿真探索决策融合诊断结果的获取。实验结果表明:神经网络和D-S证据理论相结合发挥了两者的优点,用此方法进行电梯的振动故障诊断具有较好的诊断效果。
Aiming at the shortcomings of elevator real-time fault diagnosis difficulty and low faultlocation accuracy, an information fusion diagnosis method based on neural network and D-S theory isproposed, and the corresponding fault diagnosis model is established. The principle, process andcharacteristics of wavelet packet feature parameter extraction are described and analyzed in detail, whichis applied to the decomposition and extraction of elevator car vibration signals. Taking the extractedcontent as feature vector and then combining it with neural network and D-S evidence theory, faultdiagnosis is carried out. The training process of BP neural network is further introduced, and theacquisition of decision fusion diagnosis results is explored by means of experimental simulation. Theexperimental results show that the combination of neural network and D-S evidence theory gives full playto the advantages of the two methods, and the method has a good diagnostic effect for elevator vibrationfault diagnosis.
作者
赵裕峰
ZHAO Yufeng(The Department of Information and Electrification袁 Liaoning Water Conservancy Vocational College,Shenyang 110122,China)
出处
《微处理机》
2018年第4期51-55,共5页
Microprocessors
关键词
神经网络
D-S证据理论
信息融合
特征提取
故障诊断
Neural network
D -S evidence theory
Information fusion
Feature extraction
Faultdiagnosis