Fig 1.
High-level architecture of BCQC.
Fig 2.
Flow chart for a blockchain node being involved in workpiece inspection.
Fig 3.
Flow chart for a blockchain node not involved in workpiece inspection.
Fig 4.
Flow chart for a blockchain node checking for consensus in inspection results.
Fig 5.
Floor plan of assembly shopfloor.
Table 1.
Preconditions of each step in desktop computer assembly.
Table 2.
Confusion matrix for defective workpiece detection.
Table 3.
Description of various scenarios.
Fig 6.
Total production and inspection time to assembly five desktop computers and the F1 score for each scenario with Scenario 1 as the baseline.
Fig 7.
Workpiece transport route from Workcell 15 to Workcell 2 following the direction of the conveyor belt.
Fig 8.
Distribution of QC capability between the scale of 0 to 1 for each input sets.
Fig 9.
Distribution of F1 score and stacked area of average normalized QC time across results of 20 input sets.
The QC time is the sum of transport time and inspection time. Normalization of the QC time is based on a factor of production time of the respective input sets.
Fig 10.
Equivalent threshold of majority consensus for each run in confidence consensus.
The normalized QC time (sum of transport time and inspection time) as factor of production time of the respective input sets.
Fig 11.
MSE between the estimated capability and the true capability of the trained system using Input 1.
All 17 trained input sets showed similar results.
Fig 12.
Scatter plot of normalized manufacturing time against the F1 score for all 17 inputs to manufacture five workpieces.