基于KPCA和T-S模糊神经网络的煤与瓦斯突出的预测  

Prediction of Coal and Gas Outburst Based on KPCA and T-S Fuzzy Neural Network

作　　者：顾能华 姚英彪[2] 郑慧娟 孙健[4] 王海伦 GU Neng-hua;YAO Ying-biao;ZHENG Hui-juan;SUN Jian;WANG Hai-lun(College of Electrical and Information Engineering,Quzhou University,Quzhou 324000,China;School of Communication Engineering,Hangzhou Dianzi University,Hangzhou 310018,China;Quzhou Institute of Industrial Science and Technology Information,Quzhou 324000,China;CNPC Tarim Oilfield,Korla 841000,China)

机构地区：[1]衢州学院电气与信息工程学院,浙江衢州324000 [2]杭州电子科技大学通信工程学院,浙江杭州310018 [3]衢州市工业科技信息研究所,浙江衢州324000 [4]中国石油塔里木油田,新疆库尔勒841000

出　　处：《测控技术》2018年第9期15-19,共5页Measurement & Control Technology

基　　金：国家自然科学基金资助项目(61403229)

摘　　要：应用核主成分分析(KPCA)和T-S模糊神经网络方法对煤与瓦斯突出进行快速、精准预测。利用KPCA对实验样本数据中的多种煤与瓦斯致突因素进行降维,简化问题的复杂度,将选取的累计贡献率大于90%的4个主成分作为T-S模糊神经网络的输入参数,煤与瓦斯突出强度作为输出参数。利用实测数据进行验证,并与BP神经网络预测模型、T-S模糊神经网络预测模型的预测结果进行比较。结果表明,该方法建立的预测模型准确性、有效性更高,收敛时间短,适用于煤与瓦斯突出预测。The kernel principal component analysis （KPCA） and T-S fuzzy neural network methods are used to predict coal and gas outburst rapidly and accurately. The KPCA method was used to reduce the dimension and simplify the complexity of the problem in the experimental sample data. The four principal components with the cumulative contribution rate greater than 90% were selected as T-S fuzzy neural network input parameters, coal and gas outburst strength as output. The measured data were compared with the prediction results of BP neural network prediction model and T-S fuzzy neural network prediction model under the same conditions. The results show that the predictive model established by the proposed method has higher accuracy and validity, and the convregence time is shorter, which is suitable for coal and gas outburst prediction.

关 键 词：煤与瓦斯突出 核主成分分析(KPCA) T-S模糊神经网络 仿真预测 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]