摘要
针对传统干排渣传送装置的传输距离过长,传输效率偏低,精准度差等问题。提出基于卷积神经网络的方法,解决干排渣排量问题。该方法是利用数学形态方法对元素结构进行识别,确定图像内部的隐藏形状。通过对图像边缘提取、边缘检测、池化操作、图像阈值化分析,实现干排渣传送装置故障识别。实验结果表明,基于卷积神经网络干排渣传送装置故障识别方法,与传统基于粒子群算法的干排渣传送装置故障识别方法相比,故障识别准确度提高25%,故障识别耗时缩短20~30min,故障识别准确度提高30%,故障识别耗时缩短40~50min。
In view of the problems of the traditional dry slag discharge conveying device,the transmission distance is too long,the transmission efficiency is low,and the accuracy is poor.A method based on convolutional neural network is proposed to solve the problem of dry slag discharge.This method uses the mathematical morphology method to identify the element structure and determine the hidden shape inside the image.Through the image edge extraction,edge detection,pooling operation,and image thresholding analysis,the fault identification of the dry slag conveying device is realized.The experimental results show that the fault identification method of dry slag conveying device based on convolutional neural network,compared with the traditional fault identification method of dry slag conveying device based on particle swarm algorithm,the fault identification accuracy is improved by 25%,and the fault identification time is shortened by 20%.~30min,the fault identification accuracy is increased by 30%,and the fault identification time is shortened by 40~50min.
作者
富长亮
谢晓东
张巨东
王玉玮
魏骁
Fu Changliang;Xie Xiaodong;Zhang Judong;Wang Yuwei;Wei Xiao(Beijing Guodian Futong Science and Development Co.,Ltd.Beijing 100070,China)
出处
《现代科学仪器》
2022年第5期20-24,共5页
Modern Scientific Instruments
关键词
卷积神经网络
干排渣传送装置
故障识别
网络故障识别
Convolutional Neural Network
Dry Slag Conveying Device
Fault Recognition
Network Fault Recognition