A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Prediction of Rate of Penetration Based on Random Forest in Deep Well
Abstract Rate of penetration (ROP) plays a key role in reducing drilling engineering cost. At present, low drilling rate and long drilling period have become major problems of the development of deep drilling. In response to these problems, the combination of machine learning technology and drilling engineering technology may provide new techniques for increasing the ROP. For this reason, it is necessary to introduce machine learning technology into drilling engineering, even if the work is only exploratory. In this paper, regression analysis of the ROP was conducted by the method of a variety of machine learning algorithms. In the example, a total of 15 tag data were collected, and the data volume was greater than 5,000 groups. It was saved as a CSV file and then read into a Python program. And the distribution of each tag data is analyzed, and the data distribution table of each tag data is calculated. Random forest method showed more accurate prediction result by reaching prediction accuracy of 76% in the data of whole well. In fact, if the outliers can be ignored, the accuracy will be higher.
Prediction of Rate of Penetration Based on Random Forest in Deep Well
Abstract Rate of penetration (ROP) plays a key role in reducing drilling engineering cost. At present, low drilling rate and long drilling period have become major problems of the development of deep drilling. In response to these problems, the combination of machine learning technology and drilling engineering technology may provide new techniques for increasing the ROP. For this reason, it is necessary to introduce machine learning technology into drilling engineering, even if the work is only exploratory. In this paper, regression analysis of the ROP was conducted by the method of a variety of machine learning algorithms. In the example, a total of 15 tag data were collected, and the data volume was greater than 5,000 groups. It was saved as a CSV file and then read into a Python program. And the distribution of each tag data is analyzed, and the data distribution table of each tag data is calculated. Random forest method showed more accurate prediction result by reaching prediction accuracy of 76% in the data of whole well. In fact, if the outliers can be ignored, the accuracy will be higher.
Prediction of Rate of Penetration Based on Random Forest in Deep Well
Li, Shouding (author) / Zhang, Jialiang (author) / Wu, Siyuan (author) / Chen, Weichang (author) / Chen, Dong (author) / Li, Xiao (author) / Wang, Han (author)
2019-09-25
10 pages
Article/Chapter (Book)
Electronic Resource
English
Cone Penetration Test Prediction Based on Random Forest Models and Deep Neural Networks
| Online Contents | 2023
Prediction of rate of penetration with data from adjacent well using artificial neural network
| Springer Verlag | 2019