基于改进的SVMSMOTE-Lightgbm煤与瓦斯突出预测  

Improved SVMSMOTE-Lightgbm Based Coal and Gas Outbursts Prediction

作　　者：董梦媛 郑晓亮[1,2] DONG Mengyuan;ZHENG Xiaoliang(School of Safety Science and Engineering,Anhui University of Science and Technology,Huainan Anhui 232001,China;School of Public Safety and Emergency Management,Anhui University of Science and Technology,Hefei 231100,China)

机构地区：[1]安徽理工大学安全科学与工程学院,淮南232001 [2]安徽理工大学公共安全与应急管理学院,安徽合肥231100

出　　处：《佳木斯大学学报(自然科学版)》2025年第2期5-8,16,共5页Journal of Jiamusi University:Natural Science Edition

基　　金：国家重点研发计划资助(2023YFB3211003)。

摘　　要：针对大部分煤与瓦斯预测模型所存在的预测精度低,数据样本不平衡,事故识别率低的问题,研究利用改进的SVMSMOTE算法生成数据结合Lightgbm搭建新的煤与瓦斯突出预测模型。研究结果表明,使用NRBO算法改进的SVMSMOTE算法过采样产生的数据的有效性在均值和方差上都有很好的改善,NRBO-SVMSMOTE-Lightgbm模型优于其他8个模型,改进后的SVMSMOTE-Lightgbm模型的准确率为96.22%,AUC为98%,在平均准确率方面比SVMSMOTE-Lightgbm模型提高了7.55%,提高了煤与瓦斯突出的事故识别率。基于NRBO优化的SVMSMOTE-Lightgbm模型的预测精度进一步提高,研究结果可为煤与瓦斯突出事故预测提供具体参考。Aiming at the problems of low prediction accuracy,unbalanced data samples and low accident recognition rate that exist in most coal and gas outbursts prediction models,the study utilizes the improved SVMSMOTE algorithm to generate data combined with Lightgbm to build a new coal and gas outbursts prediction model.The research results show that the validity of the data generated by oversampling using the improved SVMSMOTE algorithm of NRBO algorithm is well improved in both mean and variance,and the NRBOSVMSMOTE-Lightgbm model is better than the other eight models,and the improved SVMSMOTE-Lightgbm model has an accuracy of 96.22%and an AUC of 98%,which is better than the SVMSMOTE-Lightgbm model in terms of average accuracy.average accuracy is 7.55%higher than SVMSMOTE-Lightgbm model,which improves the accident recognition rate of coal and gas outbursts.The prediction accuracy of the SVMSMOTE-Lightgbm model based on NRBO optimization is further improved,and the results can provide a specific reference for coal and gas outbursts accident prediction.

关 键 词：煤与瓦斯突出 SVMSMOTE 数据生成 Lightgbm 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]