基于GRA-DBO-SVR的瓦斯含量预测方法  

Prediction Method of Gas Content Based on GRA-DBO-SVR

作　　者：秦宾宾 张清华[1,4] 孙国玺 张发振[4] 亢方超 李祖鹏 QIN Binbin;ZHANG Qinghua;SUN Guoxi;ZHANG Fazhen;KANG Fangchao;LI Zupeng(Guangdong Provincial Key Laboratory of Petrochemical Equipment Fault Diagnosis,Guangdong University of Petrochemical Technology,Maoming 525000,China;Postdoctoral Innovation Practice Base,Guangdong University of Petrochemical Technology,Maoming 525000,China;School of Energy and Power Engineering,Guangdong University of Petrochemical Technology,Maoming 525000,China;School of Automation,Guangdong University of Petrochemical Technology,Maoming 525000,China)

机构地区：[1]广东石油化工学院石化装备智能安全广东省重点实验室,广东茂名525000 [2]广东石油化工学院博士后创新实践基地,广东茂名525000 [3]广东石油化工学院能源与动力工程学院,广东茂名525000 [4]广东石油化工学院自动化学院,广东茂名525000

出　　处：《广东石油化工学院学报》2024年第4期80-86,共7页Journal of Guangdong University of Petrochemical Technology

基　　金：国家自然科学基金项目(61933013);茂名市科技计划项目(2024009,2024021);广东石油化工学院人才引进项目(2023rcyj2022,XJ2022000801)。

摘　　要：为提高煤层瓦斯含量预测的准确性和效率,提出了一种基于灰色关联度分析(GRA)、蜣螂优化(DBO)算法和支持向量回归(SVR)模型的瓦斯含量预测方法。采用GRA筛选影响瓦斯含量的因素来降低预测模型输入数据的维度,通过DBO算法对SVR模型的参数进行优化,构建基于GRA-DBO-SVR的瓦斯含量预测模型,并对GRA-DBO-SVR、GRA-PSO-SVR、GRA-SVR和SVR模型的预测结果进行对比。结果表明:GRA-DBO-SVR、GRA-PSO-SVR、GRA-SVR和SVR模型的MRE分别为2.82%、2.98%、3.72%和6.02%,MAE分别为0.28、0.31、0.44和0.63,MSE分别为0.17、0.18、0.37和0.90,GRA-DBO-SVR模型具有更好的泛化能力,满足工程实际需要。To improve the accuracy and efficiency of coal seam methane content prediction,a novel gas content prediction method based on Grey Relational Analysis(GRA),Dung Beetle Optimization(DBO)algorithm,and Support Vector Regression(SVR)model was proposed.First,the GRA is used to screen factors that affect gas content to reduce the dimensionality of the input data for the prediction model.Then,the DBO is employed to optimize the parameters of SVR model,constructing a gas content prediction model based on GRA-DBO-SVR.The prediction results of GRA-DBO-SVR,GRA-PSO-SVR,GRA-SVR,and SVR models are compared.The results show that the Mean Relative Errors(MRE)of GRA-DBO-SVR,GRA-PSO-SVR,GRA-SVR,and SVR are 2.82%,2.98%,3.72%,and 6.02%,respectively;the Mean Absolute Errors(MAE)are 0.28,0.31,0.44,and 0.63,respectively;and the Mean Squared Errors(MSE)are 0.17,0.18,0.37,and 0.90,respectively.The GRA-DBO-SVR model demonstrates better generalization ability,meeting the actual needs of engineering applications.

关 键 词：瓦斯含量预测 灰色关联理论 蜣螂算法 支持向量回归模型 

分 类 号：X936[环境科学与工程—安全科学]