基于PCA-BP算法的机械钻速预测研究  

Prediction for Rate of Penetration Based on PCA-BP Algorithm

作　　者：汤明 王汉昌 何世明 张光福 孔令豪 Tang Ming;Wang Hanchang;He Shiming;Zhang Guangfu;Kong Linghao(State Key Laboratory of Oil&Gas Reservoir Geology and Exploitation,Southwest Petroleum University)

机构地区：[1]西南石油大学油气藏地质及开发工程国家重点实验室

出　　处：《石油机械》2023年第10期23-31,76,共10页China Petroleum Machinery

基　　金：国家自然科学基金青年基金项目“基于水化进程的泥岩地层井壁坍塌机理研究”(51904260);国家自然科学基金面上项目“基于多场耦合理论的页岩气水平井井壁失稳机理研究”(51574202)。

摘　　要：钻井过程中,机械钻速是衡量钻井效率的一个重要指标,准确预测机械钻速对提高钻井效率、降低成本具有重要作用。常用的钻速方程预测模型存在建模困难、求解困难等问题。为此,提出一种基于主成分分析算法(PCA)优化BP神经网络的机械钻速预测新模型。基于PCA-BP模型预测机械钻速,并将预测结果与BAS-BP、BP和RF等模型的预测结果进行横向对比。结果表明,PCA-BP的拟合优度(R^(2))分别提高5.7%、9.4%和18.7%;均方根误差分别降低23.88%、30.3%和43.6%;平均绝对百分比误差分别降低35.45%、56.7%和61.5%。预测结果表明,新模型的精度更高、收敛速度更快。PCA-BP模型还可实时评价影响机械钻速因素的合理性,为提高机械钻速提供指导意见。研究结果可为实际钻进过程中提高机械钻速提供更科学的参考依据。Rate of penetration(ROP)is an important index to measure drilling efficiency.Accurate prediction of ROP plays an important role in improving drilling efficiency and reducing costs,but the commonly used ROP equation prediction model has difficulties in modeling and solving.Therefore,a new ROP prediction model based on principal component analysis algorithm(PCA)optimized BP neural network was proposed.ROP was predicted using the PCA-BP model.The predicted results were compared with those obtained from the BAS-BP,BP and RF models,showing that the goodness of fit(R^(2))of PCA-BP is increased by 5.7%,9.4%and 18.7%respectively,the root mean square error(RMSE)is decreased by 23.88%,30.3%and 43.6%respectively,and the mean absolute percentage error(MAPE)is decreased by 35.45%,56.7%and 61.5%respectively.Based on the prediction results,the new model has higher accuracy and faster convergence rate.And the PCA-BP model can also evaluate the rationality of factors affecting ROP in real time,and provide guidance for improving the ROP.The research results provide a more scientific reference for improving the ROP in the actual drilling process.

关 键 词：机械钻速预测 参数优化 小波滤波 主成分分析 BP神经网络 

分 类 号：TE21[石油与天然气工程—油气井工程]