基于AlexNet的焦炭塔工作状态识别系统研究  

Research on Working State Recognition System of Coke Tower Based on AlexNet

作　　者：谭江 王利恒[1] TAN Jiang;WANG Liheng(School of Electrical Information,Wuhan Institute of Technology,Wuhan 430205,China)

机构地区：[1]武汉工程大学电气信息学院,武汉430205

出　　处：《自动化与仪表》2025年第2期23-27,共5页Automation & Instrumentation

基　　金：武汉工程大学研究生教育创新基金项目(CX2023582)。

摘　　要：针对传统的依靠人为经验来判断焦炭塔在生焦、钻孔和除焦不同工作状态时存在工作环境恶劣、劳动强度大、主观性误差大的问题,提出了一种基于AlexNet网络模型的焦炭塔工作状态识别方法。首先,设计硬件数据采集系统采集焦炭塔在不同工作状态时的振动数据;其次,使用连续小波变换将一维振动数据转化为包含大量时频信息的二维时频图像数据,并制作时频图数据集;最后,将制作的时频图数据集作为卷积神经网络的输入,经模型训练后实现对焦炭塔工作状态的识别。实验结果表明,该方法对于焦炭塔工作状态的识别,具有识别准确率高、稳定性强等优点,识别精度能够达到91.9%,表明该方法具有可行性和有效性。In order to solve the problems of harsh working environment,high labor intensity and large subjective error in the traditional coke tower based on human experience when judging the different working states of coke tower,drilling and decoking,a method for identifying the working state of coke tower based on AlexNet network model was proposed.Firstly,a hardware data acquisition system is designed to collect the vibration data of the coke tower in different working states.Secondly,the continuous wavelet transform is used to convert the one-dimensional vibration data into two-dimensional time-frequency image data containing a large amount of time-frequency information,and the time-frequency graph dataset is made.Finally,the time-frequency graph dataset was used as the input of the convolutional neural network,and the working state of the coke tower was recognized after model training.The experimental results show that the method has the advantages of high recognition accuracy and strong stability for the identification of the working state of coke tower,and the recognition accuracy can reach 91.9%,which indicates that the method is feasible and effective.

关 键 词：数据采集系统 连续小波变换 AlexNet模型 状态识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]