3D Electrochemical Cell Imaging via Machine Learning Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing 3D images of electrochemical cell units used in microscale simulations often do not accurately represent the physical properties of the units they are intended to simulate, leading to unreliable simulations due to the use of processed or artificially generated images that may not align with the actual properties of the cells.

Innovation Solution

A computer-implemented method using supervised machine learning to label and segment 3D images of electrochemical cell units, followed by physical property calculations and adjustments to ensure accuracy, and optionally using generative machine learning to extend the image to full size, ensuring compliance with measured properties.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If 3D imaging measurement is performed on the whole electrochemical cell unit, then the image accuracy and completeness are improved, but the cost and complexity increase significantly

Engineering Contradiction:
Improveimage accuracyVSAvoidimaging complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the electrochemical cell unit into multiple portions and performs 3D imaging on each portion separately. This segmentation approach maintains measurement precision for each imaged portion while significantly reducing the overall imaging complexity and cost compared to imaging the entire cell unit at once.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent creates 3D image copies of cell portions and uses these copies for simulation purposes. By working with copied images of representative portions rather than the complete cell unit, the method achieves sufficient accuracy for simulation while avoiding the complexity of whole-cell imaging.

Inventive Principle:
Principle #26Copying

2Productivity

If processed or artificially generated images are used for simulation, then the productivity and ease of simulation are improved, but the reliability of the simulation deteriorates

Engineering Contradiction:
Improvesimulation efficiencyVSAvoidsimulation reliability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent implements a feedback mechanism where calculated physical properties from the 3D image are compared against known or measured physical properties of the electrochemical cell unit. If the calculated properties deviate from expected values, the image processing or generation parameters are adjusted iteratively to improve reliability while maintaining simulation efficiency.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent modifies parameters of the 3D image or image generation process based on physical property constraints. By changing parameters such as porosity, particle size distribution, or phase composition in the image to match known physical properties, the method ensures simulation reliability without sacrificing productivity.

Inventive Principle:
Principle #35Parameter changes

3Adaptability or versatility

If multiple configurations of electrochemical cell units are simulated, then the adaptability and design optimization are improved, but the cost and time for image acquisition increase

Engineering Contradiction:
Improveconfiguration varietyVSAvoidimage acquisition time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent creates multiple copied and modified versions of a single 3D image or a small set of reference images to represent different cell unit configurations. This copying approach enables simulation of multiple configurations without the need to acquire separate images for each configuration, thereby maintaining adaptability while minimizing time loss.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent generates different configurations by changing parameters in the 3D image or image generation process rather than acquiring new images. By modifying parameters such as porosity, particle arrangement, or phase distribution in silico, the method achieves configuration variety without additional image acquisition time.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4576013A13D images of an electrochemical cell unit consistent with measured physical properties of the electrochemical cell unit
Publication Date: 2025.06.25 TOTALENERGIES ONETECH
  • EP4576013A1 patent drawingFigure 1
  • EP4576013A1 patent drawingFigure 2~4
  • EP4576013A1 patent drawingFigure 5

AI summary

The disclosure relates to a computer-implemented method comprising: obtaining, from a 3D imaging measurement, a 3D image of at least a portion of an electrochemical cell unit; using a trained supervised machine learning engine to associate, based on intensities of voxels of the 3D image, each of said voxels of the 3D image to labels representative of electrochemical cell unit components; calculating, from said segmented 3D image, values of one or more physical property of the electrochemical cell unit; obtaining one or more physical measurements of said one or more physical property; calculating one or more difference between said one or more physical measurements and said calculated values; if said one or more difference exceeds a threshold, modifying one or more of said 3D image and one or more parameter of the preceding steps.