摘要
针对瓦斯灾害危险性预测中预测性能低的问题,对一种基于矿井内瓦斯浓度与环境因素相关性分析的瓦斯灾害选择集成预测方法进行了研究。首先,分析实验数据中样本属性与瓦斯浓度的相关性,并根据相关性分析结果进行属性约简得到新的数据集;其次,训练基学习器并应用优化集成前序选择方法建立选择集成回归学习模型;最后,将模型应用于瓦斯灾害预测。实验结果表明,基于相关性分析的选择集成回归学习模型对瓦斯灾害危险性的识别率比未进行相关性分析的四个基学习器平均提高了24%,比未进行相关性分析的选择集成回归学习模型提高了7. 6%。
In view of the low prediction performance of gas disaster risk prediction,this paper studied the prediction method of gas disaster selective ensemble regression learning based on correlation analysis of mine gas concentration and environmental factors. Firstly,this paper analyzed the correlation between gas concentration and sample attributes,and reduced the attribute to obtain a new data set according to the results of the correlation analysis. Secondly,it trained base learners,and used the optimization ensemble forward sequential selection method to establish the selective ensemble regression learning model. Finally,it used the model for gas disaster prediction. The experimental results show that the recognition rate of the proposed model for gas disaster risk compared with the four learner without correlation analysis improves 24% on average,and compared with the selective ensemble regression learning model without correlation analysis improves 7. 6%.
作者
贾澎涛
吕巧林
Jia Pengtao;Lyu Qiaolin(College of Computer Science&Technology,Xi’an University of Science&Technology,Xi’an 710054,China)
出处
《计算机应用研究》
CSCD
北大核心
2019年第12期3769-3772,共4页
Application Research of Computers
基金
西安市科技计划资助项目(2017079CG/RC042(XAKD001))
关键词
瓦斯灾害
相关性分析
选择集成回归学习
集成前序选择
识别率
gas disaster
correlation analysis
selective ensemble regression learning
ensemble forward sequential selection
recognition rate
