Downhole Signal Compression for Low-Bandwidth Surface Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current subsurface measurement techniques face challenges in accurately transmitting high volumes of data from downhole sensors to the surface due to limited data bandwidth, leading to inefficiencies in formation evaluation and drilling operations.
Innovation Solution
The system generates an augmented atom dictionary using machine-learning methods to optimize data transmission by selecting relevant signal patterns and adjusting bit allocation based on atom weights and residual characteristics, allowing for more accurate reconstruction of measurement series with reduced bit usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If mud pulse telemetry is used to transmit subsurface measurement data, then data can be communicated to the surface, but the data bandwidth is limited to a level far below the data acquisition rate
Solution Approach 1:
The patent extracts only the most essential information from the high-volume subsurface measurement data using machine learning algorithms. Instead of transmitting all acquired data, the system identifies and extracts key features and patterns that are most relevant for formation evaluation, thereby reducing the data volume to fit within the limited mud pulse telemetry bandwidth while preserving the critical information needed for accurate subsurface characterization.
Solution Approach 2:
The patent transforms the data transmission problem by changing the parameters of what is transmitted. Rather than sending raw measurement data at high volume, the system converts the data into compressed representations with optimized bit allocation. The machine learning model determines which parameters are most important and allocates more bits to those parameters, effectively changing the transmission parameters to match the bandwidth constraints while maintaining data utility.
2Loss of information
If all subsurface measurement data are transmitted to the surface, then complete information is available for analysis, but the transmission time and bandwidth requirements increase significantly
Solution Approach 1:
The patent applies preliminary action by performing data compression and feature extraction downhole before transmission. The machine learning algorithms process and compress the measurement data while it is still in the borehole, identifying and retaining only the essential information needed for formation evaluation. This preliminary processing eliminates the need to transmit unnecessary data, significantly reducing transmission time and bandwidth requirements while maintaining the completeness of critical measurement information.
3Productivity
If data compression is applied to reduce transmission bandwidth, then transmission efficiency improves, but measurement precision may be degraded
Solution Approach 1:
The patent uses parameter changes to dynamically adjust bit allocation based on the importance of different data components. The machine learning model analyzes the measurement data and identifies which parameters or features are most critical for accurate formation evaluation. More bits are allocated to these important parameters, while fewer bits are used for less critical information. This adaptive parameter adjustment maintains measurement precision for essential data while achieving significant compression for the overall data stream.
Solution Approach 2:
The patent applies local quality by treating different portions of the data with different compression levels. Rather than applying uniform compression to all data, the machine learning algorithm identifies specific regions, features, or parameters that require higher precision and preserves them with more bits, while applying more aggressive compression to other less critical portions. This localized quality preservation ensures that the most important measurement information maintains its precision even as overall data volume is reduced.
Data Source
AI summary
An apparatus includes a subsurface sensor for use in a borehole to provide a subsurface measurement series, a subsurface processor to receive the subsurface measurement series, and a machine-readable medium. The machine-readable medium has program code to cause the apparatus to obtain the subsurface measurement series and generate an atom combination based on the subsurface measurement series using the subsurface processor, wherein the atom combination comprises a subset of atoms from a dictionary. The code also has instructions to generate a set of characterizing values and transmit the set of characterizing values to a different physical location, wherein the set of characterizing values comprises an atom identifier and at least one corresponding atom weight for at least one atom from the atom combination.