Rock Classification Using CNN Microstructure Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a significant need in the oil & gas industry for an improved, computationally efficient method to assess rock properties, particularly porosity and permeability, which are crucial for estimating hydrocarbon volume and recovery potential in petroleum reservoirs.
Innovation Solution
A method involving a trained computer-implemented machine learning model, such as a convolutional neural network, processes digital images of rock microstructures to generate rock features, apply statistical processes to identify pore spaces, and classify rocks based on these features, enabling efficient assessment of rock properties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional methods are used to assess rock properties, then measurement precision can be maintained, but productivity is significantly reduced due to computational inefficiency
Solution Approach 1:
The patent applies preliminary action by pre-training a machine learning model on extensive rock microstructure datasets before actual assessment. The model learns to identify pore spaces, grains, and other microstructural features in advance, enabling rapid prediction without complex real-time computations. This pre-computation approach maintains measurement precision while dramatically improving productivity during actual rock property assessment.
Solution Approach 2:
The patent replaces traditional mechanical/image processing methods with a machine learning-based system. Instead of using conventional algorithms to analyze rock microstructure images, the system employs trained neural networks that can rapidly process images and predict rock properties. This substitution maintains accuracy while achieving significant computational efficiency improvements.
2Measurement precision
If detailed microstructure analysis is performed to improve measurement precision, then device complexity increases due to the need for advanced processing systems
Solution Approach 1:
The patent applies self-service by designing a machine learning model that automatically performs feature extraction, pore space identification, and rock property prediction without requiring complex post-processing systems. The model self-adjusts during training and independently handles the entire analysis pipeline, reducing the need for additional complex processing equipment and software layers.
3Measurement precision
If multiple imaging scales are used to improve measurement precision, then loss of time increases due to processing multiple images
Solution Approach 1:
The patent merges multiple imaging scale analyses into a single integrated machine learning model. The model is trained to process and analyze features across different scales simultaneously, combining information from various magnifications into unified rock property predictions. This merging approach maintains the measurement precision benefits of multi-scale analysis while reducing total processing time by eliminating sequential analysis steps.
Data Source
AI summary
A method, apparatus, and program product perform microstructure analysis of a digital image of rock using a trained convolutional neural network model to generate a plurality of rock features. The rock features can represent a pore space in the microstructure of the rock including pores and throats. In many implementations, a statistical process can be applied to the rock features to generate characteristics of the pore space which can be used in classifying the rock.