基于深度自编码器的城市固废焚烧过程燃烧状态识别  

Combustion State Recognition of Municipal Solid Waste Incineration Process Based on Deep Autoencoder

作　　者：王印松[1] 赵佳玉 WANG Yinsong;ZHAO Jiayu(Department of Automation,North China Electric Power University,Baoding 071003,China)

机构地区：[1]华北电力大学自动化系,河北保定071003

出　　处：《控制工程》2024年第10期1729-1737,共9页Control Engineering of China

摘　　要：国内城市固废(municipal solid waste,MSW)的组分复杂且多变,其焚烧过程的燃烧状态识别主要依靠人工判断,难以维持稳定的运行工况。针对上述问题,提出了一种基于深度自编码器的分区域燃烧状态识别方法。首先,依据炉排结构对燃烧段和燃烬段的分界线进行标定;然后,利用具有深层结构的卷积稀疏自编码器(convolutional sparse autoencoder,CSAE)提取两部分火焰图像的特征;最后,将特征分别输入到相应的最小二乘支持向量机进行状态识别。基于处理规模为750 t/d的焚烧炉的不同燃烧状态图像进行实验,实验结果表明,在有限的标记样本数量下,所提方法的平均识别准确率为98.04%,该方法能够实现MSW燃烧状态的实时监测。The composition of municipal solid waste(MSW)is complex and variable.The combustion state recognition in the incineration process of MSW mainly depends on manual judgment,which is difficult to maintain a stable operating condition.To solve the above problem,a regional combustion state recognition method based on deep autoencoder is proposed.Firstly,a demarcation line between the burning section and the burnout section is demarcated according to the grate construction.Then,convolutional sparse autoencoder(CSAE)with a deep structure is used to extract features from two parts of flame images.Finally,the features are input to respective least square support vector machine(LSSVM)for state recognition.Based on different combustion state images of 750 t/d incinerators,the experimental results show that the average recognition accuracy of the proposed method is 98.04%under the limited number of labeled samples,and the method can realize real-time monitoring of MSW combustion state.

关 键 词：炉排炉 燃烧状态识别 深度自编码器 最小二乘支持向量机 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]