机器学习预测机械钻速及在工程上的应用

机械钻速是钻井工程中的关键指标，直接关系到钻井成本，准确判断机械钻速是钻井工程决策的关键。针对传统机械钻速预测方法误差大、时效低的问题，尝试通过机器学习的方法获取更准确、可靠的机械钻速预测模型。首先采用不同的机器学习算法初步建立机械钻速预测模型，然后通过性能比选优选出梯度提升树算法，最终通过优化参数建立机械钻速预测模型。将训练好的预测模型用于南海某盆地一口探井，预测结果符合实际。采用模型预测的结果可以用于机械钻速评价，识别机械钻速异常，为工程决策提供依据。

Rate of penetration (ROP) is a key performance indicator in drilling engineering, which is directly related to drilling cost. The accurate judgment of ROP is the key of drilling engineering decision. To solve the problems of large error and low effective of traditional ROP prediction methods, this paper attempts to obtain a more accurate and reliable ROP prediction model through machine learning method. Different machine learning methods are used to preliminarily establish the prediction model of ROP, and then the gradient lifting tree algorithm is selected for predicting ROP through performance comparison. Finally, the prediction model of ROP is established by optimizing parameters. The trained prediction model is applied to an exploration well in a basin in South China Sea, and the prediction results are in line with the reality. The results predicted by the model can be used to evaluate the ROP, identify the abnormal ROP, and provide a basis for engineering decision-making.