A drilling correction prediction method and system based on LSTM

Through the LSTM-based drilling correction prediction method, multi-source data preprocessing and model training were used to solve the problem of deviation in the drilling process, achieve higher drilling correction accuracy and efficiency, and improve the accuracy of deep hole construction.

CN120336734BActive Publication Date: 2025-09-12POWERCHINA RAILWAY CONSTR +3
2 Cites 0 Cited by

Patent Information

Application Number
CN202510796966.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-06-16
Publication Date
2025-09-12
Estimated Expiration
2045-06-16

AI Technical Summary

Technical Problem

The existing technology causes drilling deviations during the drilling process due to the complexity and uncertainty of geological conditions. The existing multi-source data has not achieved the fusion of temporal and spatial features. The traditional machine learning model cannot effectively capture the coupling effect of the gradual change of formation inclination and drill tool wear, resulting in increased deviations in the middle and late stages of drilling, which seriously affects the accuracy of deep hole construction.

Method used

A drilling correction prediction method based on LSTM is adopted. By acquiring and preprocessing historical multi-source data, a drilling posture deviation prediction model is constructed. The powerful memory and nonlinear fitting capabilities of LSTM are utilized to capture the time series characteristics of the drilling process. Combined with multi-segment speed change information fusion, multi-directional joint search and position cross transformation methods, the model training effect is improved.

Benefits of technology

It improves the accuracy and efficiency of drilling correction, can more accurately capture the time series characteristics of the drilling process, and effectively improves the accuracy of drilling construction.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention discloses a drilling correction prediction method and system based on LSTM, which belongs to the field of data recognition technology. A drilling posture deviation prediction model is constructed by LSTM, and the drilling posture deviation prediction model is trained by using training data to obtain the drilling posture deviation prediction model after training, and then real-time multi-source data is collected and normalized. The real-time multi-source data after normalization is input into the trained drilling posture deviation prediction model for identification to obtain a drilling posture deviation prediction result, and finally the drilling posture deviation prediction result is used as a drilling correction prediction result, and the drilling correction prediction result is transmitted to a device designated by a staff member, so that the staff can perform drilling correction. By utilizing the powerful memory capacity and nonlinear fitting capacity of LSTM, the time series characteristics of the drilling process can be more accurately captured, and the drilling correction efficiency and accuracy can be effectively improved.
Need to check novelty before this filing date? Find Prior Art