摘要
基于神经网络系统在采煤机故障检测中的应用,以某矿MG-300/700-WD采煤机截割机构故障诊断系统为例,简单介绍了监测数据准备与本文所用神经网络模型结构建立过程,通过比较不同匹配阈值下神经网络训练过程中达到允许误差所需训练次数,以及评估故障诊断效果,对神经网络故障检测模型结构中匹配阈值q的取值进行优化选择,确定了最优匹配阈值为q=0.43。研究结果确保了采煤机截割减速器箱故障预测及诊断的准确性,提高工作面生产效率。
Based on the application of neural network system in the fault detection of continuous mining machine,the process of monitoring data preparation and the neural network model structure established in this paper is briefly introduced.By comparing the training times required to reach the allowable error in the neural network training process under different matching thresholds.The fault diagnosis effect is evaluated,and the value of the matching threshold q in the neural network fault detection model structure is optimized,and the optimal matching threshold is determined to be q=0.43.The research results ensure the accuracy of fault prediction and diagnosis of the cutting gearbox of the continuous mining machine and improve the production efficiency of the working face.
作者
李佳杰
Li Jiajie(Kaiyuan Coal Industry Co.^Ltd.,Shouyang Shanxi 045400)
出处
《机械管理开发》
2020年第3期119-121,共3页
Mechanical Management and Development
关键词
神经网络系统
故障检测
匹配阈值
优化
neural network system
fault detection
matching threshold
optimization