基于CNN-LSTM算法的电梯振动故障预测  

Prediction of Elevator Vibration Fault Based on CNN-LSTM Algorithm

作　　者：艾学忠[1] 张玉龙 徐春博 Ai Xuezhong;Zhang Yulong;Xu Chunbo(School of Information and Control Engineering,Jilin Institute of Chemical Technology,Jilin,Jilin 132022,China)

机构地区：[1]吉林化工学院信息与控制工程学院,吉林吉林132022

出　　处：《机电工程技术》2025年第4期171-176,共6页Mechanical & Electrical Engineering Technology

摘　　要：为了克服传统电梯故障检测系统在精度和动态范围上的不足,在硬件层面采取了将高性能STM32微控制器与MPU6050加速度计相结合的技术方案,目的是实现对电梯振动数据的高精度采集。在软件层面采用了卷积神经网络-长短期记忆网络(CNNLSTM)的深度学习算法模型,对采集到的电梯振动数据信号进行深入分析,有效识别出与电梯故障相关的关键特征,并进行准确的预测分析。通过该智能分析系统,可以实时监控电梯的运行状态,并在上位机界面直观地展示预测结果。结果显示,该系统能够拟合电梯振动信号的整体波动趋势,在不排除外界人为干扰的情况下,预测结果能够达到83%,且预测集整体损失值为0.000 6。该系统能够很好的适应电梯运行环境,对电梯运行状态实时检测,而且能够提前识别出潜在的故障信息。In order to overcome the shortcomings of the traditional elevator fault detection system in accuracy and dynamic range,a technical scheme combining a high-performance STM32 microcontroller and an MPU6050 accelerometer has been adopted at the hardware level,aiming at realizing high-precision acquisition of elevator vibration data.At the software level,the deep learning algorithm model of convolutional neural network-long short yerm memory network(CNN-LSTM)is adopted to conduct in-depth analysis of the collected elevator vibration data signals,effectively identify the key features related to elevator faults,and carry out accurate prediction analysis.Through the intelligent analysis system,the running state of the elevator can be monitored in real time,and the prediction results can be displayed intuitively on the upper computer interface.The results show that the system can fit the overall fluctuation trend of the elevator vibration signal,and the prediction result can reach 83%without excluding the external human interference,and the overall loss value of the prediction set is 0.0006.The system can well adapt to the operating environment of the elevator,detect the operating state of the elevator in real time,and identify the potential fault information in advance.

关 键 词：电梯振动信号 故障预测 CNN-LSTM算法 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TU857[自动化与计算机技术—控制科学与工程]