Skip to main content

Table 2 Mean and standard deviation (in italics) of RMSE scores (best performing in bold) in the DTW-graph and correlation-graph for all traffic occurrences for every algorithm

From: A computational framework for modeling complex sensor network data using graph signal processing and graph neural networks in structural health monitoring

Algorithm

Sensor type

 

Non-filtered

Filtered

 

X-strain

Y-strain

Combined

X-strain

Y-strain

Combined

Correlation

Random

0.80

1.36

1.12

0.45

0.86

0.68

 

(.32)

(.95)

(.76)

(.29)

(.62)

(.46)

Top-down

0.60

1.06

0.74

0.31

0.66

0.38

 

(.24)

(.75)

(.52)

(.19)

(.51)

(.29)

Bottom-up

0.68

1.08

0.88

0.34

0.71

0.46

 

(.30)

(.80)

(.68)

(.21)

(.53)

(.35)

DTW

      

Random

1.07

0.98

1.09

0.47

0.50

0.48

 

(.66)

(.34)

(.41)

(.31)

(.22)

(.22)

Top-down

0.76

0.76

0.64

0.31

0.37

0.29

 

(.38)

(.24)

(.27)

(.20)

(.16)

(.16)

Bottom-up

0.99

0.90

0.83

0.42

0.42

0.35

 

(.65)

(.29)

(.37)

(.32)

(.16)

(.18)

  1. The columns non-filtered and filtered show whether or not graph signal filtering was used