Scholarlink

<p>This paper presents a smart system for detecting faults in single-phase induction motors. It uses real-time monitoring and machine learning on an affordable embedded platform. The system architecture incorporates a Raspberry Pi 4 and two modules of ADS1115 analog-to-digital converter, interfaced with a suite of sensors which includes an ADXL335 triaxial accelerometer for vibration analysis, SCT-013-030 current sensor, DS18B20 digital temperature sensor, and an AC voltage sensor. All this captures critical motor health indicators. Real-time data from five motor fault conditions which are front-end bearing fault, winding short circuit, open capacitor fault, undervoltage, and normal operation was used to train and evaluate multiple machine learning models. Random Forest was the best classifier, achieving an accuracy of 98.5%. It also had a precision of 97.65%, recall of 97.32%, F1-score of 97.47%, and a ROC-AUC of 99.90%. Models like Naive Bayes and KNN showed strong results too. This proves the dataset is robust. BLYNK IOT platform has been used in this system to provide visual aid of parameters and machine health status can also be checked through it. The proposed system offers a scalable,cost-effective, and accurate way to predict faults for single phase induction motor.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Intelligent automated fault detection framework for single phase motors using real time monitoring and machine learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月1日

查看全文
  • Abhinab Shukla,
  • S. P. Shukla,
  • Saji Chacko,
  • Abhishek Kumar Tripathi,
  • Taoufik Saidani,
  • N. Rao Cheepurupalli,
  • Yewuhalashet Fissha

摘要

This paper presents a smart system for detecting faults in single-phase induction motors. It uses real-time monitoring and machine learning on an affordable embedded platform. The system architecture incorporates a Raspberry Pi 4 and two modules of ADS1115 analog-to-digital converter, interfaced with a suite of sensors which includes an ADXL335 triaxial accelerometer for vibration analysis, SCT-013-030 current sensor, DS18B20 digital temperature sensor, and an AC voltage sensor. All this captures critical motor health indicators. Real-time data from five motor fault conditions which are front-end bearing fault, winding short circuit, open capacitor fault, undervoltage, and normal operation was used to train and evaluate multiple machine learning models. Random Forest was the best classifier, achieving an accuracy of 98.5%. It also had a precision of 97.65%, recall of 97.32%, F1-score of 97.47%, and a ROC-AUC of 99.90%. Models like Naive Bayes and KNN showed strong results too. This proves the dataset is robust. BLYNK IOT platform has been used in this system to provide visual aid of parameters and machine health status can also be checked through it. The proposed system offers a scalable,cost-effective, and accurate way to predict faults for single phase induction motor.