Innovative Application of Drilling Engineering Data Quality Control Based on Generative Artificial Intelligence Technology

Abstract

Abstract Innovative achievements of quality inspection in data government science applications based on digital technology will provide data guarantee for the construction of oil and gas production and operations. To provide intelligent application support for the quality inspection of massive and diverse data formats in drilling and completion engineering, this article intends to explore the intelligent parsing of unstructured text, as well as the judgment of quality control index rules based on parsed information. This article explores the technical solution for extracting unstructured document information in drilling and completion engineering, and proposes and implements a set of unstructured data quality control methods that integrate intelligent parsing and consistency verification. Firstly, an intelligent parsing framework for multi type engineering tables was constructed, which comprehensively utilizes PDF parsing, and regular expressions to accurately extract wellbore master data and engineering parameters from design documents. A set of end-to-end structured information extraction system based on RapidOCR (PP-OCRv4 Mobile) and large language model (Qwen3:1.7B) was designed for the comprehensive logging map in image format. Through dynamic field extraction and intelligent text restoration, high-precision extraction of key quality control data items in complex layout drawings was achieved. By automatically comparing the extracted structured information with authoritative sources of enterprise master data, the system achieves cross document and cross-source data consistency verification. The experiment was validated based on a dataset of 114 multi format logging comprehensive maps, and the system achieved an average field level accuracy of 88.89%. The average single file processing time was about 66.28 seconds, effectively demonstrating the feasibility and efficiency of the proposed method. The intelligent question answering system Textwiz for data quality control is developed. This method provides an efficient and intelligent solution for quality control of unstructured documents. The application results show that the professional document Q&A system constructed using Retrieval Augmented Generation (RAG) technology can be applied in scenarios such as logging completion reports, logging blueprint information extraction, quality inspection, and exploration and development professional knowledge Q&A. This method can replace manual and implement automated document quality inspection, with an accuracy rate of 95%. It provides the practicality and effectiveness of intelligent question-answering systems in unstructured data quality inspection.