摘要
煤层开采过程中的工作面矿压分析与预测,对煤矿顶板管理与安全生产具有重要意义。然而,工作面开采引起的围岩移动和变形影响着矿压预测的准确度。为了提高工作面来压位置预测的精度,以分布式光纤监测采动覆岩变形的频移数据为基础,引入门控循环神经网络(gated recurrent neural networks, GRU),建立了遗传算法(genetic algorithm, GA)-GRU-反向传播(back propagation, BP)的工作面来压位置预测模型。将光纤频移值的统计特征融合工作面推进距离等因素作为特征向量,并采用GA对GRU及BP网络的超参数寻优。实验结果表明:预测模型的决定系数为98.7%,平均绝对误差为1.224 cm,均方根误差为1.769 cm,预测的准确性高,为工作面矿压预测提供了新的方法。
The analysis and prediction of the mine pressure has important practical significance for the management of the roof and safe mining production in a coal mine. However, during the continuous advancement of the working face, the deformation and movement of surrounding rock affects the accuracy of rock pressure prediction seriously. In order to improve the accuracy of predicting pressure position at the working face, a genetic algorithm-gated recurrent neural networks-back propagation(GA-GRU-BP) model was proposed by introducing the GRU. It was based on the frequency shift value data of distributed optical fiber monitoring mining overburden deformation. The statistical characteristics of the frequency shift value on the optical fiber sensor, working face advance distance and other factors affecting the mine pressure were used as the eigenvectors. The GA was used to support super-parametric optimization of GRU and BP neural network. The results show that the proposed model has a high accuracy, which the coefficient of determination of the proposed model is 98.7%, the average absolute error is 1.224 cm, and the root mean square error is 1.769 cm. A new method is provided for predicting the mine pressure on the working face.
作者
冀汶莉
田忠
张丁丁
欧阳一博
JI Wen-li;TIAN Zhong;ZHANG Ding-ding;OUYANG Yi-bo(School of Communication and Information Engineering,Xi'an University of Science and Technology,Xi'an 710054,China;School of Energy,Xi'an University of Science and Technology,Xi'an 710054,China)
出处
《科学技术与工程》
北大核心
2022年第24期10485-10492,共8页
Science Technology and Engineering
基金
国家重点研发计划(2018YFC0808301)
国家自然科学基金(51804244)。
关键词
工作面矿压
工作面来压位置预测
GA-GRU-BP
光纤频移值
mine pressure of working face
pressure position prediction on working face
genetic algorithm-gated recurrent neural networks-back propagation(GA-GRU-BP)
frequency shift value
作者简介
第一作者:冀汶莉(1973-),女,汉族,陕西西安人,硕士,副教授。研究方向:机器学习及煤矿信息化。E-mail:jiwenli@xust.edu.cn。
