基于极限学习机的矿井突水水源快速识别模型

发布时间：2018-06-29 05:21

本文选题：矿井突水 + 水源识别　；参考：《煤炭学报》2017年09期

【摘要】：在煤矿突水灾害防治过程中,需要快速准确地识别出突水水源类型。激光诱导荧光技术具有灵敏度高和快速监测的特点,利用该技术获取水样的荧光光谱。光谱经卷积平滑预处理和主成分分析提取特征信息后,采用极限学习机算法建立多元分类学习模型。确定隐含层激励函数为Sigmoid函数,并通过交叉验证法确定最优隐含层节点个数。从训练网络的平均时间、训练和测试的平均分类准确率和标准差方面,与BP和SVM传统分类算法进行了性能比较。结果表明:在训练集和测试集上的平均分类准确率方面,该模型与传统分类模型基本一致,但该模型分类准确率的标准差最小,说明其具有较稳定的分类性能;在训练模型学习时间方面,该模型能够大幅度降低分类学习时间,说明其具备快速识别突水水源性能。
[Abstract]:In the process of prevention and treatment of water inrush disaster in coal mine, it is necessary to identify the type of water inrush water source quickly and accurately. Laser induced fluorescence technique has the characteristics of high sensitivity and rapid monitoring. After the spectral information was extracted by convolution smoothing and principal component analysis (PCA), the multivariate classification learning model was established by extreme learning machine (LLM) algorithm. The hidden layer excitation function is determined as Sigmoid function and the optimal number of hidden layer nodes is determined by cross-validation. The average time of training network, the average classification accuracy and standard deviation of training and testing are compared with the traditional classification algorithms of BP and SVM. The results show that the model is basically consistent with the traditional classification model in terms of the average classification accuracy on the training set and the test set, but the standard deviation of the classification accuracy of the model is the minimum, which indicates that the model has stable classification performance. In terms of learning time of training model, the model can greatly reduce the learning time of classification, which shows that the model has the capability of identifying water inrush water sources quickly.
【作者单位】：安徽理工大学电气与信息工程学院;阜阳师范学院计算机与信息工程学院;中国矿业大学资源与地球科学学院;淮南矿业集团谢桥煤矿;
【基金】：“十二五”国家科技支撑计划资助项目(2013BAK06B01) 国家安全生产重大事故防治关键技术科技资助项目(anhui-0001-2016AQ) 安徽省自然科学研究基金资助项目(KJ2015A278)
【分类号】：TD745.2

【参考文献】