孙元田, 李桂臣, 钱德雨, 张苏辉, 许嘉徽. 巷道松软煤体流变参数反演的BAS-ESVM模型与应用[J]. 煤炭学报, 2021, 46(S1): 106-115. DOI: 10.13225/j.cnki.jccs.2020.0878
引用本文: 孙元田, 李桂臣, 钱德雨, 张苏辉, 许嘉徽. 巷道松软煤体流变参数反演的BAS-ESVM模型与应用[J]. 煤炭学报, 2021, 46(S1): 106-115. DOI: 10.13225/j.cnki.jccs.2020.0878
SUN Yuantian, LI Guichen, QIAN Deyu, ZHANG Suhui, XU Jiahui. Research and application of BAS-ESVM model for rheological parameter inversion of soft coal mass in roadway[J]. Journal of China Coal Society, 2021, 46(S1): 106-115. DOI: 10.13225/j.cnki.jccs.2020.0878
Citation: SUN Yuantian, LI Guichen, QIAN Deyu, ZHANG Suhui, XU Jiahui. Research and application of BAS-ESVM model for rheological parameter inversion of soft coal mass in roadway[J]. Journal of China Coal Society, 2021, 46(S1): 106-115. DOI: 10.13225/j.cnki.jccs.2020.0878

巷道松软煤体流变参数反演的BAS-ESVM模型与应用

Research and application of BAS-ESVM model for rheological parameter inversion of soft coal mass in roadway

  • 摘要: 深部煤体巷道变形具有明显的流变特性,准确高效地获取煤体的流变参数是研究巷道流变机理的重要基础。基于典型的煤巷流变工程案例,经分析认为,巷道帮部的松软煤体的长时变形具有两阶段流变特征,即前期的减速大流变阶段和后期的等速大流变阶段,获得煤体的流变参数对于进一步研究巷道失稳具有重要意义。基于人工智能算法支持向量机(SVM),采用天牛须算法(BAS)高效获取SVM的核函数参数σ和罚参数C,形成改进型进化支持向量机(ESVM),提高SVM的学习和泛化能力。进一步地,明确围岩的流变模型和弹塑性力学参数,基于正交设计原理,构建松散煤体流变参数样本数据。建立数值模拟模型,对构建的流变参数样本进行模拟计算,得到含有时序性的计算巷道位移,进而将每一组流变参数及其通过数值计算得到的与时间相对应的两帮移近量位移作为一组样本,形成样本数据库。将上述数据通过ESVM模型不断学习,以现场位移数据为目标,再利用BAS搜索得到最佳的煤体流变参数,最终构建得到BAS-ESVM煤体流变参数反演模型。得到的流变参数经过正算验证,结果显示计算位移与实测值吻合较好,证明了该方法的有效性与流变参数的正确性,该方法为研究煤体巷道流变机理奠定了参数基础。

     

    Abstract: The deformation of deep coal roadway has rheological characteristics.Accurate and efficient acquisition of rheological parameters of coal is an important basis for the study of rheological mechanism.Based on a typical case of rheological coal roadway, it is considered that the long-term deformation of the soft coal mass in the side wall of the roadway has two-stage rheological characteristics, i.e.the deceleration stage in the early stage and the constant velocity stage in the later stage.Obtaining the rheological parameters of the coal mass is of great significance for a further analysis of roadway instability.The artificial intelligence algorithms, Support Vector Machine(SVM) and Beetle Antennae Search Algorithm(BAS) are used.Aiming at the problem of selecting the kernel function parameter and penalty parameter C of SVM accurately, the improved evolutionary Support Vector Machine(ESVM) is composed to improve the learning and generalization ability of SVM by using the efficient optimization characteristics of BAS.Furthermore, the rheological model and elastic-plastic mechanical parameters of surrounding rock mass are defined.Based on the principle of orthogonal design, the dataset of rheological parameters of loose coal are constructed.A numerical model is established to model the deformation related to rheological parameters, and the calculated roadway displacement with time sequence is obtained.Then each group of rheological parameters and the displacement of sidewalls corresponding to time obtained by numerical calculation are taken as a group of samples to form a database.The above data are continuously learned through the ESVM model, and the best rheological parameters of coal mass are obtained by using BAS.Finally, the BAS-ESVM inversion model of coal rheological parameters is established.The obtained rheological parameters are verified by forward calculation.The results show that the calculated displacement is in good agreement with the measured values.The validity of the method and the correctness of the rheological parameters are proved.The method lays a parameter foundation for the study of the rheological mechanism of coal roadway.

     

/

返回文章
返回