摘要
结合智慧矿山的应用背景,针对MD580-70×9这一类型的矿用水泵,从故障判断、故障诊断以及信号预测三个方面出发,希望通过模拟水泵振动信号得到一种较为普适的应用模型,在时域利用模拟水泵进行故障的判断以及基于长短时记忆(LSTM)神经网络对时域信号进行预测,在判断出故障后通过FFT进行时频转换,进而判断出水泵故障类型,实现对矿用水泵进行检测。
Combined with the application background of smart mines,for the MD580-70×9 type of mine pump,from the three aspects of fault judgment,fault diagnosis and signal prediction,this paper hopes to obtain a kind of more general application model by simulating the vibration signal of the water pump.It uses the analog water pump in the time domain to judge the fault and predicts the time domain signal based on the Long-Short Time Memory (LSTM) neural network.It carries out the time-frequency conversion through FFT after judging the fault,and then judges the type of water pump fault,realizes the detection of mine water pump.
作者
谢正卿
邱磊
杨丽晴
袁志浩
崔钰豪
XIE Zhengqing;QIU Lei;YANG Liqing;YUAN Zhihao;CUI Yuhao(China University of Mining and Technology-Beijing,Beijing 100083,China)
出处
《现代信息科技》
2022年第13期155-157,161,共4页
Modern Information Technology
关键词
水泵故障检测
深度学习
长短期记忆网络
water pump fault detection
deep learning
Long-Short Term Memory network
