基于模糊聚类和SVM的瓦斯涌出量预测  

Gas Emission Quantity Forecasting Based on Fuzzy Clustering and SVM

作　　者：王晓路[1] 

机构地区：[1]西安科技大学通信与信息工程学院,西安710054

出　　处：《煤炭技术》2011年第9期93-94,98,共3页Coal Technology

基　　金：电子信息产业发展基金招标项目(XDJ2-0514-27);国家高技术研究发展计划(863计划)(2005AA133070);陕西省教育厅项目(09JC05)

摘　　要：为了准确预测瓦斯涌出量,提出了一种基于模糊聚类和支持向量机(SVM)的瓦斯涌出量预测方法。将瓦斯涌出量相关影响因素作为特征空间中的样本,采用模糊C均值聚类对特征空间中的样本进行聚类分析,对于所得到的不同类别样本分别建立SVM预测模型。结果表明:采用单纯的SVM预测方法,对于不同特征的样本的预测个别预测误差相对较大,其最大误差为8.11%,平均误差为4.68%,采用文中所建议的用FCM对样本分类后再进行SVM预测,预测精度有明显改善,最大误差和6.94%,平均误差为3.35%,表明所建议的方法是有效和可行的。In order to accurately predict the gas emission quantity,an approach based on fuzzy clustering and support vector machine（SVM） is proposed.The factors related gas emission are used as the sample in characteristics space.The sample is conducted clustering analysis by fuzzy c-means clustering（FCM）,the different classified samples are obtained,based on which,the corresponding SVM based forecasters is constructed,respectively.The results show that simply implanted SVM based forecasters some deviations are relatively large for the different future samples and the averaged prediction biases is 4.68% and the maximum deviation is 8.11%.The precision of the prediction is obviously improved to obtain the averaged biases of 3.35% and the maximum error of 6.914% by using the SVM based forecaster based on the proposed the clustering samples obtained by FCM.It is indicated that the suggested approach is feasible and effective.

关 键 词：瓦斯涌出量 模糊C均值聚类 SVM 预测 

分 类 号：TD713[矿业工程—矿井通风与安全]