封闭储煤场煤自燃高温点运移规律及反演预测研究  

Study on transportation law and inverse prediction of high-temperature points during spontaneous combustion of coal in closed coal storage yard

作　　者：张嬿妮 姚迪[1,2] 张陆陆 舒盼 段正肖 翟芳妍 ZHANG Yanni;YAO Di;ZHANG Lulu;SHU Pan;DUAN Zhengxiao;ZHAI Fangyan(College of Safety Science and Engineering,Xi’an University of Science and Technology,Xi’an Shaanxi 710054,China;Shaanxi Key Laboratory of Prevention and Control of Coal Fire,Xi’an University of Science and Technology,Xi’an Shaanxi 710054,China;Key Laboratory of Coal Resources Exploration and Comprehensive Utilization,Ministry of Land and Resources,Xi’an Shaanxi 710021,China)

机构地区：[1]西安科技大学安全科学与工程学院,陕西西安710054 [2]西安科技大学陕西省煤火灾害防治重点实验室,陕西西安710054 [3]国土资源部煤炭资源勘查与综合利用重点实验室,陕西西安710021

出　　处：《中国安全生产科学技术》2024年第6期78-84,共7页Journal of Safety Science and Technology

基　　金：国家自然科学基金项目(52174199);陕西省杰出青年科学基金项目(2023-JC-JQ-46)。

摘　　要：为探究封闭储煤场煤自燃高温点运移规律及反演预测方法,通过搭建煤自燃模拟实验台研究封闭储煤场煤高温点运移规律,分析煤自燃表层与内部温度数据之间的相关性,采用遗传算法优化前后的随机森林(RF)、BP神经网络(BPNN),建立封闭储煤场煤自燃高温点预测模型。研究结果表明:松散煤自燃过程中,高温点运移方向呈非线性移动规律,其主要受煤燃烧过程中裂隙作用影响;表层位点与下部各层之间的温度数据呈现较强相关性;预测结果中RF和BPNN均具有较强鲁棒性和容错率,BPNN优秀的非线性映射能力,使得其处理结果优于RF;对RF和BPNN进行优化后,遗传算法的优化对RF和BPNN预测准确率均有提升,对RF影响最大,模型优化后精度达到98%以上。研究结果可为封闭储煤场煤自然发火防治提供参考。In order to investigate the transportation law and inverse prediction method of high-temperature points during the spontaneous combustion of coal in closed coal storage yard,the transportation law of high-temperature points of coal in closed coal storage yard was studied by constructing a simulation experimental bench of coal spontaneous combustion,and the correlation between the surface layer and internal temperature data of coal spontaneous combustion was analyzed.The random forest(RF)and BP neural network(BPNN)before and after the optimization of genetic algorithm were used to establish a prediction model of high-temperature points during the spontaneous combustion of coal in closed coal storage yard.The results show that during the spontaneous combustion process of loose coal,the transportation direction of high-temperature points presents a nonlinear moving law,which is mainly affected by the effect of fissure in the process of coal combustion.The temperature data between the surface layer sites and the lower layers presents the strong correlation.Both RF and BPNN have strong robustness and fault tolerance in the prediction results,and the excellent nonlinear mapping ability of BPNN makes its processing results better than RF.After the optimization of RF and BPNN,the optimization of genetic algorithm improves the prediction accuracy of both RF and BPNN,and the impact on RF is the greatest,with the accuracy of model reaching more than 0.98 after optimization.The research results can provide a reference for the prevention and control of natural ignition of coal in the closed coal storage yards.

关 键 词：封闭储煤场 煤自燃 温度预测 智能算法 遗传算法 

分 类 号：X932[环境科学与工程—安全科学]