基于自动化机器学习的故障诊断系统研究

Research on fault diagnosis system based on automated machine learning

针对泵站监测系统缺少状态关系分析和故障诊断系统功能扩展性差的问题,文章研究实现了可自动构建诊断算法的故障诊断系统。在故障诊断领域,样本容易出现不平衡问题。为解决该问题,文章探索研究了样本平衡性检验方法、样本不平衡下算法训练和效果验证方法。通过研究基于遗传算法的自动化机器学习实现该功能,系统可以方便地实现功能扩展与更新。

In response to the lack of state relationship analysis and poor functional scalability of the fault diagnosis system in the pump station monitoring system,this paper studies and implements a fault diagnosis system that can automatically construct diagnostic algorithms.In the field of fault diagnosis,samples are prone to imbalanced issues.To address this issue,explore and study methods for sample balance testing,algorithm training under sample imbalance,and effectiveness verification.By studying automated machine learning based on genetic algorithms to achieve this function,the system can easily expand and update its functions.