Digital Twinning remote laboratories for online practical learning

Figures & data

Figure 1. Collaboration with the VR learning environment.

Figure 2. Digital twin builder architecture.

Table 1. Digital twin platform datasources.

Data source	Data type	Displays	Examples	Uses
Design data	csv	Graphs/Bar Charts	Simulation Data	Performance, Trend Analysis and Limits
	2D Data	3D plots	Response Surface	Performance, Trend Analysis and Limits
	CAD	3D Object	As-Designed Geometry	Review
Manufacturing data	csv	Graphs/Bar Charts	Machine tool performance, consumable levels	Performance, Trend Analysis and Limits
	IoT	Graphs/Bar Charts	Machine tool performance, consumable levels	Performance, Trend Analysis and Limits – ordering of consumables
	Mesh	3D Objects	Design for Manufacture Geometry	Comparison of as-designed and as-made
	Point Clouds	3D Objects	As-Made Geometry	Comparison of as-designed and as-made, Review
Test data	csv	Graphs/Bar Charts	System performance	Performance, Trend Analysis and Limits
	IoT	Object Texturing	Sensor Data	Performance, Trend Analysis and Limits
	Mesh	Object Texturing	Test Article	Review
Service data	csv	Graphs/Bar Charts	In-Service performance, i.e. utilisation rates	Performance, Trend Analysis and Limits – Data Analytics
	IoT	Graphs/Bar Charts, Object Texturing	Real time system configuration data	Performance, Trend Analysis and Limits – Data Analytics
	Mesh	Object Texturing	Scanned Objects	Review, Performance, Trend Analysis and Limits – Data Analytics

Figure 3. Digital twin data pipeline.

Figure 4. Overall architecture of the geometry pipeline. The ‘Reduce’ steps highlighted are expanded upon in .Figure 5

Figure 5. Geometry pipeline ‘reduce’ step. Copies of this process are run in parallel for every unique part in the model. The three components highlighted in blue are the three machine learning models.

Table 2. Shape and similarity metrics used in the geometry pipeline to ensure mesh visual quality. Note that none of these metrics encode the scale of the object, to ensure the method is scale invariant.

Symbol	Metric	Type
L1	Bounding box length/width ratio	Shape Metric
L2	Bounding box length/depth ratio	Shape Metric
φ	Sphericity	Shape Metric
ρBox	Bounding box density (mesh volume/bounding box volume)	Shape Metric
E	Shape Efficiency E = RM/RB	Shape Metric
Askew	Skewness in polygon area	Shape Metric
Akurt	Kurtosis in polygon area	Shape Metric
ACOV	Coefficient of variance in polygon area	Shape Metric
α	Mesh connectivity (ratio of polygons to vertices in mesh)	Shape Metric
RV	Low poly/High poly volume ratio	Shape Ratio
RA	Low poly/High poly surface area ratio	Shape Ratio
Rφ	Low poly/High poly sphericity ratio	Shape Ratio
Rρ	Low poly/High poly bounding box density ratio	Shape Ratio
RE	Low poly/High poly shape efficiency ratio	Shape Ratio
Rα	Low poly/High poly mesh connectivity ratio	Shape Ratio
Rskew	Low poly/High poly polygon area skewness ratio	Shape Ratio
Rkurt	Low poly/High poly polygon area kurtosis ratio	Shape Ratio
RCOV	Low poly/High poly polygon area coefficient of variance ratio	Shape Ratio
D	Average distance between equivalent points	Similarity Metric
∆N	Average normal vector deviation	Similarity Metric
∆θ	Average dihedral angle deviation	Similarity Metric

Figure 6. Mesh quality training data collection application (original on the left, processed on the right).

Figure 7. Example model produced with the geometry pipeline. The casing has been rendered transparent to show the internal details.

Figure 8. Overview of process pipeline model.

Figure 9. Hierarchy of an FBX file.

Figure 10. Modified FBX hierarchy used within the scenario definition.

Figure 11. The scenario definition file link between the digital twin builder and player.

Figure 12. The case study brief.

Figure 13. Specifying the case study within the digital twin builder.

Figure 14. The electrical laboratory within the digital twin player.

Figure 15. The digital twin player showing an equipment item together with its data simulation results.