Probabilistic Data Fusion for Real-Time Bit-Rock Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for rig state detection during drilling operations lack efficiency in characterizing the relationship between the drill bit and rock, leading to suboptimal drilling performance and increased operational risks.
Innovation Solution
A system utilizing a probabilistic mixture model, detection engine for trends, and network model for inference, which acquires and analyzes data to characterize the bit-rock interaction, enabling real-time monitoring and adjustment of drilling operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional rig state detection methods are used, then the system is simpler to implement, but the measurement precision of bit-rock interaction characterization is insufficient
Solution Approach 1:
The patent segments the complex data analysis task into three distinct modules: a probabilistic mixture model for identifying operational modes, a detection engine for extracting trends, and a network model for inference. This segmentation allows each module to specialize in a specific aspect of bit-rock interaction characterization, improving overall measurement precision while managing system complexity through modular architecture.
Solution Approach 2:
The patent introduces intermediate processing layers between raw data acquisition and final inference. The probabilistic mixture model acts as an intermediary that transforms raw sensor data into identified modes, which then feed into the detection engine for trend extraction, and finally into the network model for inference. These intermediaries enhance characterization accuracy by progressively refining the data at each stage.
2Productivity
If real-time data analysis is implemented, then drilling efficiency is improved, but the computational resources and time required increase
Solution Approach 1:
The probabilistic mixture model performs preliminary classification of operational modes before detailed analysis. By pre-identifying the current operational mode (e.g., rolling, sliding, bouncing), the system can apply mode-specific analysis parameters and reduce computational overhead for subsequent trend detection and inference, enabling real-time processing without sacrificing accuracy.
Solution Approach 2:
The system dynamically adjusts its analysis approach based on the identified operational mode. Different drilling modes (rolling, sliding, bouncing) have different characteristic signatures, and the system adapts its detection engine parameters and network model inputs according to the current mode, optimizing computational efficiency for each specific condition while maintaining real-time responsiveness.
3Measurement precision
If multiple data types are analyzed, then the inference accuracy is improved, but the device complexity increases
Solution Approach 1:
The network model serves as a universal inference engine that handles multiple data types and operational modes through a unified probabilistic framework. Rather than implementing separate analysis systems for different data types, the single network model is trained to process diverse inputs (vibration, torque, pressure, etc.) and infer bit-rock interactions across all operational conditions, reducing overall system complexity while maintaining high inference accuracy.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method includes acquiring data during rig operations where the rig operations include operations that utilize a bit to drill rock and where the data include different types of data; analyzing the data utilizing a probabilistic mixture model for modes, a detection engine for trends and a network model for an inference based at least in part on at least one of a mode and a trend; and outputting information as to the inference where the inference characterizes a relationship between the bit and the rock