摘要
针对燃煤机组锅炉系统故障检测效能不足的现状,根据系统多参量高度耦合的特征,本文采用多尺度特征融合策略构建智能分析架构,开发了一种结合深度特征学习的智能化诊断框架,利用卷积神经网络的层级抽象能力实现故障信号的非线性映射与模式解耦。实验结果表明:提出模型具有93.2203%准确率,该方法在故障判别方面展现出显著优势,为锅炉可靠性运行提供有力保障。
ion ability of convolutional neural networks is utilized to achieve the non-linear mapping and pattern decoupling of fault signals.Experimental results show that the proposed model has an accuracy of 93.2203%.This method demonstrates significant advantages in fault discrimination,providing a strong guarantee for the reliable operation of boilers.
作者
管晨希
任永灿
张松松
王林
GUAN Chen-xi;REN Yong-can;ZHANG Song-song;WANG Lin(China National Coal Group Corp.,Beijing 100120,China;Shenyang University of Chemical Technology,Shenyang 110142,China;China Special Equipment Inspection Group Co.,Ltd.,Beijing 100029,China)
出处
《节能技术》
2025年第4期317-321,共5页
Energy Conservation Technology
关键词
锅炉
特征分析
运行参数
故障诊断
神经网络
boiler
characteristic analysis
operating parameters
fault diagnosis
neural network
