Li Dongyang, Yang Jianyi, Pan Zaisheng, Li Nanyang
College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou 310013, China.
Hangzhou Special Equipment Inspection and Research Institute, Hangzhou 310051, China.
Sensors (Basel). 2023 Jul 24;23(14):6646. doi: 10.3390/s23146646.
It is important to improve the identification accuracy of the operating status of elevator traction machines. The distribution difference of the time-frequency signals utilized to identify operating circumstances is modest, making it difficult to extract features from the vibration signals of traction machines under various operating conditions, leading to low recognition accuracy. A novel method for identifying the operating status of traction machines based on signal demodulation method and convolutional neural network (CNN) is proposed. The original vibration time-frequency signals are demodulated by the demodulation method based on time-frequency analysis and principal component analysis (DPCA). Firstly, the signal demodulation method based on principal component analysis is used to extract the modulation features of the experimentally measured vibration signals. Then, The CNN is used for feature vector extraction, and the training model is obtained through multiple iterations to achieve automatic recognition of the running state. The experimental results show that the proposed method can effectively extract feature parameters under different states. The diagnostic accuracy is up to 96.94%, which is about 16.61% higher than conventional methods. It provides a feasible solution for identifying the operating status of elevator traction machines.
提高电梯曳引机运行状态的识别精度至关重要。用于识别运行情况的时频信号分布差异不大，使得在各种运行条件下从曳引机振动信号中提取特征变得困难，导致识别精度较低。提出了一种基于信号解调方法和卷积神经网络（CNN）的曳引机运行状态识别新方法。通过基于时频分析和主成分分析（DPCA）的解调方法对原始振动时频信号进行解调。首先，利用基于主成分分析的信号解调方法提取实测振动信号的调制特征。然后，使用CNN进行特征向量提取，并通过多次迭代获得训练模型以实现对运行状态的自动识别。实验结果表明，该方法能够有效提取不同状态下的特征参数。诊断准确率高达96.94%，比传统方法高出约16.61%。它为识别电梯曳引机的运行状态提供了一种可行的解决方案。
