Towards the Automation of Infrared Thermography Inspections for Industrial Maintenance Applications

. 2022 Jan 13;22(2):613. doi: 10.3390/s22020613

Algorithm 1 An algorithm proposed to develop a robust and fast predictive maintenance system based on thermographic images

1:
Select industrial equipment
2:
Generate thermographic labeled dataset (reference + abnormal conditions)
3:
process_time ← thermographic data processing
4:
ROI candidates ← importance analysis
Ensure:
Sort ROIs in descending order of importance
5:
while metric((N+1) $_{R O I s}$ ) < metric(N $_{R O I s}$ )do
6:
$N_{R O I s} \leftarrow$ (N+1) $_{R O I s}$
7:
end while
8:
for feature in features do
9:
metric(feature)
10:
end for
11:
$f e a t u r e \leftarrow$ max(metric(feature))
12:
for model in models do
13:
train_time(model)
14:
prediction_time(model)
15:
metric(model)
16:
end for
17:
$m o d e l \leftarrow$ optimize(process_time, train_time, prediction_time, metric)