嵌入式传感器结合机器学习的振动模式分类方法  

Vibration Pattern Classification Method Combining Embedded Sensors with Machine Learning

作　　者：李佳俊 钱嵩橙 李林行 Li Jiajun;Qian Songcheng;Li Linhang(School of Automation Engineering,University of Electronic Science and Technology of China,Chengdu 611731,China;Xingzhi College,Chengdu College of University of Electronic Science and Technology of China,Chengdu 611731,China;School of Computer Science and Engineering,Sichuan University of Science and Engineering,Yibin,Sichuan 644000,China)

机构地区：[1]电子科技大学自动化工程学院,成都611731 [2]电子科技大学成都学院行知学院,成都611731 [3]四川轻化工大学计算机科学与工程学院,四川宜宾644000

出　　处：《机电工程技术》2024年第11期220-223,共4页Mechanical & Electrical Engineering Technology

摘　　要：开发一种结合STM32L432KC微控制器和LIS3DH加速度传感器的振动数据智能分析方法,用以诊断机械设备故障。通过频域滤波技术优化信号处理,并采用随机森林算法进行故障模式的识别与分类,该方法显著提升了振动模式检测的准确性和效率。实验结果证明,该方法能够有效区分不同振动模式,并表现出较高的识别准确率,体现了嵌入式系统与机器学习算法相结合在异常振动检测中的优越性,对预防工业安全事故和降低维护成本具有显著意义。此外,该研究还探讨了振动信号处理和模式分类的技术细节,包括信号采集、预处理、特征提取与分类算法的选择和优化,旨在提供一种高效、准确的故障诊断工具。该研究成果不仅为机械故障诊断提供了一种新的技术手段,也为未来嵌入式系统在工业应用中的研究提供了新的思路和方向。A method for intelligent vibration data analysis is developed by integrating an STM32L432KC microcontroller and LIS3DH accelerometer to diagnose mechanical equipment failures.By optimizing signal processing through frequency domain filtering and employing a random forest algorithm for fault mode recognition and classification,the method significantly enhances the accuracy and efficiency of vibration mode detection.Experimental results confirm the method’s effectiveness in distinguishing different vibration modes,demonstrating a high recognition accuracy rate.The discovery underscores the superiority of combining embedded systems with machine learning algorithms in abnormal vibration detection,significantly impacting industrial safety incident prevention and maintenance cost reduction.Additionally,the study explores technical details of vibration signal processing and mode classification,including signal acquisition,preprocessing,feature extraction,and the selection and optimization of classification algorithms,aiming to provide an efficient and accurate diagnostic tool for mechanical fault diagnosis.The findings not only offer a new technical approach for mechanical fault diagnostics but also pave new avenues for future research on embedded systems in industrial applications.

关 键 词：嵌入式 模式识别 机器学习 随机森林算法 故障诊断 

分 类 号：TP368.2[自动化与计算机技术—计算机系统结构]