摘要
为提高煤与瓦斯突出预测的准确率和效率,提出了一种基于数据预处理的多策略改进烟花算法(IFWA)优化极限学习机(ELM)的煤与瓦斯突出预测模型。首先,针对于非线性多维特征数据,使用灰色关联度分析(GRA)进行特征选取,利用主成分分析(PCA)进行特征约简,将数据预处理后的数据指标作为模型的输入;其次,引入引力搜索算子和混合变异策略改进烟花算法(FWA)易陷入局部最优的问题;最后,利用IFWA对ELM的输入层到隐含层的权重和偏差进行优化,构建最优的煤与瓦斯突出风险预测模型。结果表明,IFWA-ELM模型的RMSE和R2可达0.074,0.968,与ELM、GA-ELM、PSO-ELM和FWA-ELM模型相比均有所提升,IFWA-ELM模型对煤与瓦斯突出危险等级预测的准确率可达100%,具有更好的收敛速度和预测精度。研究成果可为煤矿瓦斯突出多数据融合预测提供可靠的理论依据。
In order to improve the accuracy and efficiency of coal and gas outburst prediction,a coal and gas outburst prediction model based on data preprocessing multi-strategy improved fireworks algorithm(IFWA)optimized extreme learning machine(ELM)was proposed.Firstly,for the nonlinear multi-dimensional feature data,the grey relational analysis(GRA)was used for feature selection,the principal component analysis(PCA)was used for feature reduction,and the data index after data preprocessing was used as the input of the model.Secondly,the gravitational search operator and hybrid mutation strategy were introduced to improve the problem that the fireworks algorithm(FWA)was easy to fall into local optimum.Finally,IFWA was used to optimize the weight and deviation from the input layer to the hidden layer of ELM,and the optimal coal and gas outburst risk prediction model was constructed.The results show that the RMSE and R~2 of IFWA-ELM model can reach 0.074 and 0.968,which are improved compared with ELM,GA-ELM,PSO-ELM and FWA-ELM models.The accuracy of prediction of IFWA-ELM model on coal and gas outburst risk level can reach 100%,which has better convergence speed and prediction accuracy.The research results can provide a reliable theoretical basis for multi-data fusion prediction of coal mine gas outburst.
作者
乔威豪
安葳鹏
赵雪菡
吕常周
QIAO Weihao;AN Weipeng;ZHAO Xuehan;LYU Changzhou(School of Computer Science and Technology,Henan Polytechnic University,Jiaozuo,Henan 454000,China;School of Software,Henan Polytechnic University,Jiaozuo,Henan 454000,China)
出处
《矿业研究与开发》
CAS
北大核心
2024年第5期98-105,共8页
Mining Research and Development
基金
国家自然科学基金项目(61872126)
河南省重点科技攻关资助项目(192102210123)。
关键词
煤与瓦斯突出
烟花算法
极限学习机
数据预处理
风险预测模型
Coal and gas outburst
Fireworks algorithm
Extreme learning machine
Data preprocessing
Risk prediction model
