期刊文献+

基于IsoMap和MBFO-SVR的瓦斯涌出量动态预测研究 被引量:4

Based on the IsoMap with MBFO-SVR Gas Emission Dynamic Prediction Research
下载PDF
导出
摘要 为了能够实现高精度与实时性的动态预测煤矿绝对瓦斯涌出量,本文提出了等容特征映射Iso Map(Isometric feature Mapping)与改进细菌觅食优化算法MBFO(Modified Bacteria Foraging Optimization)优化支持向量回归机SVR(Support Vector Regression)相结合的预测方法。瓦斯涌出是在多种影响因子共同作用下的结果,并且这些因素之间是复杂的非线性关系,因此本文中提出采用流形学习方法 Iso Map对其进行降维特征提取,该方法用测地距离(geodesic distace)取代了普遍采用的欧氏距离,有利于对高维特征内在关系的挖掘,取得了优于传统的主成分分析(PCA)的结果;将MBFO算法对SVR的相关参数进行寻优;将Iso Map分析结果输入预测模型。仿真表明,与PSO算法比较,本文提出的预测方法预测精度较高,更加有利于对瓦斯涌出量预测。 In order to realize the dynamic prediction of absolute gas emission with high precision and real time in coal mine,this paper puts forward a forecasting method by combining the Isometric feature Mapping(IsoMap)and Support Vector Regression machine(SVR)optimized byModified Bacteria Foraging algorithm(MBFO). Gas emission is an emergent property resulting from various interactions,and these factors are complex nonlinear relationship. Therefore,using the IsoMap,a manifold learning method,is to reduce the dimension of feature extraction in this article. This methodis advantageous to excavate the high dimensioneigenvectorinner relationship by using geodesic distanceto replace the Euclidean distanceand superior to the traditional principal component analysis(PCA);By using MBFO to optimizing parameters of SVR,results analysised by IsoMap are the input of prediction model. Simulation shows that compared with PSO algorithm,the proposed prediction method forecasting accuracy is higher, more conducive to the quantity of gas emission prediction.
出处 《传感技术学报》 CAS CSCD 北大核心 2016年第7期1115-1120,共6页 Chinese Journal of Sensors and Actuators
基金 国家自然科学基金项目(51274118) 辽宁省教育厅基金项目(UPRP20140464)
关键词 瓦斯涌出量 等容特征映射 细菌觅食优化算法 支持向量回归机 gas emission Isometric feature mapping bacteria foraging optimization support vector regression machine
  • 相关文献

参考文献22

二级参考文献260

共引文献407

同被引文献47

引证文献4

二级引证文献26

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部