基于万有引力的煤层底板突水预测算法

发布时间：2019-01-03 06:16

【摘要】：传统的基于机器学习的煤层突水预测方法需要大量的训练样本进行预测模型的训练。而在矿井突水问题中,大量训练样本的获得基本上是不可能的。主要研究在突水样本数据有限的情况下提高煤层突水预测结果的准确性。结合山西省某煤矿的实际情况,提出了一种新颖的基于万有引力的煤层底板突水预测方法(Gravitational force based algorithm,GFA)。该算法采用半监督的学习方式,将万有引力公式引入到预测模型中,利用少量的突水训练样本作为引力的源点吸引测试样本进行突水安全状态的传递,进而实现突水测试样本安全性的预测。将提出的算法用于历史突水数据以及实际的煤层底板突水数据进行实验,实验结果表明,在突水训练数据有限的情况下,提出的基于万有引力的煤层底板突水预测算法可获得良好的预测效果。
[Abstract]:The traditional prediction method of coal seam water inrush based on machine learning needs a large number of training samples to train the prediction model. However, in the problem of mine water inrush, it is basically impossible to obtain a large number of training samples. This paper mainly studies how to improve the accuracy of water inrush prediction in coal seam under the condition of limited water inrush sample data. Based on the actual situation of a coal mine in Shanxi Province, a novel method of predicting water inrush from coal seam floor based on gravity is presented in this paper, (Gravitational force based algorithm,GFA). The algorithm uses semi-supervised learning method to introduce the universal gravity formula into the prediction model and uses a small number of water inrush training samples as the source of gravity to attract the test samples to transfer the safe state of water inrush. Furthermore, the safety prediction of water inrush test samples is realized. The proposed algorithm is applied to the experiment of historical water inrush data and actual water inrush data of coal seam floor. The experimental results show that, under the condition of limited water inrush training data, The prediction algorithm of water inrush from coal seam floor based on gravitation can obtain good prediction results.
【作者单位】：太原理工大学信息工程学院;
【基金】：山西省重大专项基金资助项目(20130321004-01) 山西省科技攻关资助项目(20121101004)) 教育部博士后基金资助项目(2013M530896)
【分类号】：TD745

【参考文献】