基于机器学习的镐型截齿截割力预测

Prediction of cutting force acting on the conical pick based on machine learning

截齿的截割力是衡量其能否实现破岩的重要参数,对掘进机械的设计选型至关重要。本文基于截割破岩试验数据,构建包含训练集和测试集的截割力特征集,研究了不同机器学习方法的镐型截齿截割力预测能力。采用均方误差和决定系数作为模型预测能力的评估指标,研究表明,多项式回归模型的预测模型效果最好,其性能优于传统计算模型,最适合用于建立截割力预测模型;线性回归类模型对截割力预测性能较差,不适合用于建立截割力预测模型;决策树类模型的性能相对于多项式模型略差,也有一定的应用前景。本研究为截齿截割力的预测提供了新的解决思路。

The cutting force of pick is an important parameter to measure whether it can achieve rock breaking,which is crucial to the design and selection of tunneling machinery.Based on the rock cutting test data,the cutting force feature set including training set and test set was constructed,and the cutting force prediction ability of the conical pick with different machine learning methods was studied.Mean square error and coefficient of determination were adopted as indicators for assessing the predictive power of models.The study shows that the polynomial regression model has the best predictive effect,which outperforms the traditional computational model and it is most suitable for modelling the cutting force prediction.The linear regression model has poor performance for cutting force prediction,which are not suitable for modelling the cutting force prediction.The decision tree model has slightly worse performance relative to the polynomial model,but it also has some application prospects.This study provides a new solution for the prediction of the cutting force acting on the conical pick.