期刊文献+

基于SAPSO-BP的CO_(2)相变致裂效果预测及敏感度分析

Prediction of CO_(2)Phase Transformation Cracking Effect and Sensitivity Analysis Based on SAPSO-BP
下载PDF
导出
摘要 液态CO_(2)相变致裂效果受很多因素影响,针对BP神经网络收敛速度慢且易陷入局部最优解的问题,为提高模型的预测精度和泛化能力,采用SAPSO算法优化BP神经网络的权值和阈值,并采用MATLAB软件编写构建了SAPSO-BP算法,基于仿真得到的35组数据,使用SAPSO-BP,PSO-BP,BP模型以及多元线性回归模型对致裂效果进行预测,其平均相对误差分别为2.73%、7.1%、14.6%、13.9%。平均绝对误差分别为0.068、0.169、0.239、0.314 m。表明:SAPSO-BP算法预测精度最高,提高了BP模型的预测精度,其精度满足工程实际需要。并采用Sobol指数法探究了相关影响因素对有效致裂半径的敏感度,表明:敏感度由高到低依次为地应力、弹性模量、瓦斯压力、泄放压力、致裂器间距、钻孔直径、抗拉强度,可为CO_(2)相变致裂的工程设计提供理论支持。 The cracking effect of liquid CO_(2)phase change is affected by many factors.The convergence speed of BP neural network is slow and easy to fall into the problem of local optimal solution.In order to improve the prediction accuracy and generalization ability of the model,SAPSO algorithm is used to optimize the weight and threshold of BP neural network,and SAPSO-BP algorithm is compiled and constructed by MATLAB software.Based on 35 groups of data obtained from simulation,SAPSO-BP,PSO-BP the average relative errors of BP model and multiple linear regression model are 2.73%,7.1%,14.6%and 13.9%respectively.The average absolute errors are 0.068,0.169,0.239 and 0.314 m respectively.It shows that SAPSO-BP algorithm has the highest prediction accuracy,improves the prediction accuracy of BP model,and its accuracy meets the actual needs of engineering.The sensitivity of relevant influencing factors to the effective cracking radius is explored by Sobol index method.It shows that the sensitivity from high to low is:in-situ stress,elastic modulus,gas pressure,relief pressure,crack spacing,borehole diameter and tensile strength,which can provide theoretical support for the engineering design of CO_(2)phase change cracking.
作者 张增辉 王长禄 邢迎欢 ZHANG Zenghui;WANG Changlu;XING Yinghuan(Baode Coal Mine,China Shenhua Shendong Coal Group,Xinzhou 036600,China;China Coal Research Institute,Beijing,100013,China;College of Safety Science and Engineering,Liaoning Technical University,Fuxin 123000,China;Key Laboratory of Mine Power Disaster and Prevention of Ministry of Education,Liaoning Technical University,Huludao 125105,China)
出处 《煤炭技术》 CAS 北大核心 2023年第4期172-177,共6页 Coal Technology
关键词 CO_(2)相变致裂效果预测 退火粒子群算法 SAPSO-BP神经网络 MATLAB Sobol指数法 CO_(2)phase transformation cracking effect prediction annealing particle swarm optimization SAPSO-BP neural network MATLAB Sobol index method
  • 相关文献

参考文献19

二级参考文献167

共引文献272

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部