摘要
燃煤炉选煤重介分选过程中,采集到的煤泥浆含量、磁物质含量等数据可能存在误差,无法准确捕捉到煤炭的关键特征,导致重介分选过程无法克服意外扰动,控制只能采用反馈形式,控制效果较差。为此,提出基于卷积神经网络前馈补偿的选煤厂燃煤炉重介质分选智能控制技术。通过仪器实时采集选煤工艺介质中的煤泥浆含量、重介质密度和磁物质含量等分选参数。通过卷积神经网络模型识别评价煤泥滤饼含水比例,作为重介分选智能控制依据;基于历史数据和煤泥饼含水量数据,设计前馈补偿方法。通过补偿精煤灰分分选中的扰动,得到精煤灰分的分选模型,估算出合理的分选参数数值。将通过仪器采集的含量参数作为前馈输入特征,根据实时监测和估算出的分选参数数值,对输出期望数值进行动态调整。实验结果表明,方法对燃煤炉选煤重介分选过程智能控制效果好,煤泥含水量波动较小,能保证选煤效率和质量。
In the process of coal selection and heavy medium separation in coal-fired furnaces,there may be errors in the collected data such as coal slurry content and magnetic material content,which cannot accurately capture the key characteristics of coal.As a result,the heavy medium separation process cannot overcome unexpected disturbances,and control can only be completed in the form of feedback later,resulting in poor control effect.To this end,a convolutional neural network-based feedforward compensation based intelligent control technology for heavy medium sorting in coal-fired boilers of coal preparation plants is proposed.Real time collection of separation parameters such as coal slurry content,heavy medium density,and magnetic material content in the coal selection process medium through instruments.Using convolutional neural network models to identify and evaluate the water content ratio of coal slurry filter cakes,as a basis for intelligent control of heavy medium sorting;Design a feedforward compensation method based on historical data and coal slurry cake moisture content data.By compensating for the disturbance caused by the selection of clean coal ash,a sorting model for clean coal ash is obtained,and reasonable sorting parameter values are estimated.The content parameters collected through the instrument will be used as feedforward input features,and the expected output values will be dynamically adjusted based on the real-time monitoring and estimation of sorting parameter values.The experimental results show that the proposed method has a good intelligent control effect on the coal selection and heavy medium separation process of the combustion furnace,with small fluctuations in the moisture content of the coal slurry,and can ensure the efficiency and quality of coal selection.
作者
刘军
LIU Jun(Guoneng Shendong Coal Group Washing Center,Shenmu 719315,China)
出处
《洁净煤技术》
CAS
CSCD
北大核心
2024年第S01期577-582,共6页
Clean Coal Technology
关键词
前馈补偿
卷积神经网络
燃煤炉
选煤重介分选
介质密度
feedforward compensation
convolutional neural networks
coal fired boilers
coal selection and heavy medium separation
medium density