基于随机森林的矿压预测方法  

Mine pressure prediction method based on random forest

作　　者：冀汶莉[1,2] 刘艺欣 柴敬 王斌[1] JI Wenli;LIU Yixin;CHAI Jing;WANG Bin(College of Communication and Information Engineering,Xi'an University of Science and Technology,Xi'an 710054,China;Key Laboratory of Western Mine Exploitation and Hazard Prevention,Ministry of Education,Xi'an University of Science and Technology,Xi'an 710054,China;College of Energy Engineering,Xi'an University of Science and Technology,Xi'an 710054,China)

机构地区：[1]西安科技大学通信与信息工程学院,陕西西安710054 [2]西安科技大学西部矿井开采及灾害防治教育部重点实验室,陕西西安710054 [3]西安科技大学能源学院,陕西西安710054

出　　处：《采矿与岩层控制工程学报》2021年第3期67-77,共11页Journal of Mining and Strata Control Engineering

基　　金：国家重点研发计划资助项目(2018YFC0808301);国家自然科学基金资助项目(41027002,51804244)。

摘　　要：煤矿开采过程中工作面矿压显现的分析与预测,对于保障煤矿安全生产具有重要意义。分布式光纤监测技术是煤矿开采过程中覆岩变形监测的新方法,以分布式光纤监测采动覆岩变形相似物理模拟试验为背景,建立了基于随机森林的MBCT-SR-RF工作面来压预测模型。首先提出光纤加权频移平均变化度,并引入多步逆向云变换算法(MBCT-SR)计算光纤上所有测点频移数据的期望Ex、熵En和超熵He等统计特征;然后以光纤加权频移平均变化度和光纤频移数据的统计特征(Ex,En,He)作为输入样本,以均方根误差(RMSE)、平均绝对误差(MAE)和平均绝对百分比误差(MAPE)作为性能评估指标;最后进行了预测模型的泛化能力研究。试验结果表明,基于MBCT-SR-RF的工作面来压预测方法的RMSE为5.2896 cm,MAE为4.3367 cm,MAPE为3.9167%,与BP神经网络和SVM支持向量机方法相比,均低于相应的评价指标,具有较高的准确率、鲁棒性和泛化能力。该方法实现了基于光纤频移数据的工作面来压预测,为煤矿开采过程中的顶板智能化管理提供了判断依据。The analysis and prediction of mineral pressure manifestation in the coal mining process are of great significance to guarantee the safe and efficient production of coal mines.Distributed optical fiber monitoring technology has become a new method for monitoring overlying rock deformation during coal mining.This thesis takes the physical similarity simulation test of distributed optical fiber monitoring overburden deformation as the research background,and establishes the MBCT-SR-RF mineral pressure manifestation prediction model based on random forest.Firstly,the concept of fiber weighted frequency shift average change degree is defined,and introduce the multi-step backward cloud transform(MBCT-SR)to calculate the expected(Ex),entropy(En),and hyper-entropy(He)of the frequency shift data of all measurement points on the fiber.And then,the fiber weighted frequency shift average change degree and the fiber frequency shift data statistical features(Ex,En,He)are used as the feature attributes of the input samples.Root mean square error(RMSE),average absolute error(MAE)and average absolute percentage error(MAPE)are used as performance evaluation metrics.The experimental results show that RMSE,MAE and MAPE are respectively 5.2896 cm,4.3367 cm and 3.9167%.Compared with the BP neural network and the SVM support vector machine method,the prediction method of the pressure position of the working face based on MBCT-SR-RF has higher accuracy and robustness.This method realizes the mineral pressure manifestation using optical fiber frequency shift data.It provides a judgment basis for intelligent roof management during the mining process of the working face.

关 键 词：矿压显现 来压位置预测 MBCT-SR 随机森林 分布式光纤 

分 类 号：TD324[矿业工程—矿井建设]