摘要
采用经极限学习机训练的神经网络建立故障诊断模型,基于该模型设计了一种矿井主要通风机故障诊断系统,介绍了该系统的软硬件设计方案。测试结果表明,该系统中极限学习机算法运行时间仅为0.031 3s,故障诊断准确率不低于97.35%,其实时性和准确性优于基于BP神经网络、ELMAN神经网络、经支持向量机训练的神经网络等模型的主要通风机故障诊断系统。
A fault diagnosis model was built by use of neural network trained by extreme learning machine. A fault diagnosis system of mine main ventilator based on the model was designed, and software and hardware design schemes of the system were introduced. The test results show running time of extreme learning machine algorithm in the system is only 0. 031 3 s and accuracy rate of fault diagnosis is not less than 97.35 %, which has better real-time performance and accuracy than fault diagnosis systems based on BP neural network, ELMAN neural network or neural network trained by support vector machine.
出处
《工矿自动化》
北大核心
2017年第6期69-71,共3页
Journal Of Mine Automation
基金
国家自然科学基金资助项目(61303183)
关键词
煤炭开采
主要通风机
故障监测
故障诊断
极限学习机
coal mining
main ventilator
fault monitoring
fault diagnosis
extreme learning machine