Gradient Boosting 算法在典型浅埋煤层液压支架选型中的应用

    Application of Gradient Boosting Algorithm in Hydraulic Support Selection of Typical Shallow Coal Seam

    • 摘要: 针对目前工作面液压支架阻力确定方法的不足,提出了1种新的预测方法,采用改进后的逻辑斯提算法(LR)来优化梯度提升回归(GBRT)模型,以此来预测液压支架阻力。在GBRT中加入学习速率来限制子模型的学习速率,防止其过拟合;应用LR对样本参数进行优化,建立LR-GBRT回归预测模型;将该预测模型应用于液压支架阻力的预测,预测结果与LR(线性回归模型)、SVM(支持向量机模型)、DTR(决策树回归模型)、EN(弹性网回归模型)进行对比分析。结果表明:LR-GBRT模型具有较强的泛化能力,较高的预测精度,可以对液压支架阻力进行有效预测。

       

      Abstract: A new method using gradient boosting regression tree(GBRT) which is optimized by logistic regression(LR) to predict working resistance of hydraulic support is proposed, avoiding shortcomings of current determining methods. Add learning rate to GBRT to limit the learning rate of sub-models and prevent over-fitting; using LR to optimize sample parameters to establish LR-GBRT regression prediction model; the model is applied to predict resistance of hydraulic support, and the prediction result is compared with linear regression(LR), support vector model(SVM), decision tree regression(DTR), elastic net regression(EN). The results show that the model has better generalization performance and higher prediction accuracy. It can effectively predict resistance of hydraulic support

       

    /

    返回文章
    返回