含油污泥催化热解特性与预测

Characteristics and prediction for catalytic pyrolysis of oil sludge

采用热重分析仪研究了含油污泥(Oily Sludge,OS)的催化热解特性,并基于人工神经网络(Artificial Neural Network,ANN)算法,建立了可预测催化效率的机器学习模型,为催化热解过程优化提供指导。研究表明Fe_(2)O_(3)、CaO可降低OS在2个热解阶段的热失重峰值温度,且OS第1热解阶段的热失重峰值温度均由315℃降至270℃,第2热解阶段的热失重峰值温度由460℃分别降至420和430℃。反应动力学模型表明Fe_(2)O_(3)、CaO可降低OS有机组分的热解活化能,由33 kJ/mol分别降至25和30 kJ/mol,Fe_(2)O_(3)催化效率更高。建立的机器学习模型可准确预测催化热解过程(R^(2)=0.99,R^(MSE)=0.03)。基于机器学习模型预测,选用Fe_(2)O_(3)作为热解催化剂,混合比13%,热解温度440℃时,OS热解催化效率最高,可达10.27%,本研究将为OS等有机固废的催化热解提供新思路。

This research investigated the catalytic pyrolysis characteristics of oily sludge(OS)using athermogravimetric analyzer.Based on the Artificial Neural Network(ANN)algorithm,a machine learning model predicting catalytic efficiency was established to guide the optimization of the catalytic pyrolysis process.The study indicates that Fe_(2)O_(3)and CaO can decrease the peak thermogravimetric temperature of OS.The peak thermogravimetric temperature decreases from 315℃to 270℃at the first pyrolytic stage,and decreases from 460℃to 420 and 430℃,respectively,at the second pyrolytic stage.The reaction kinetic models suggest that Fe_(2)O_(3)and CaO can reduce the pyrolysis activation energy of organic components in OS from 33 kJ/mol to 25 kJ/mol and 30 kJ/mol,respectively,with Fe_(2)O_(3)exhibiting a higher catalytic efficiency.The established machine learning model accurately predicts the catalytic pyrolysis process(R^(2)=0.99,R^(MSE)=0.03).According to the prediction of machine learning model,when choosing Fe_(2)O_(3)as the pyrolysis catalyst with a mixing ratio of 13%at a pyrolysis temperature of 440℃,the catalytic efficiency of OS can reach its peak at 10.27%.This study provides a new perspective for the catalytic pyrolysis of organic solid waste like OS.