Table 8 Test results of computing cost of SMA-SVM model on large-scale real engine data set.
From: Fault diagnosis using ISMA to optimize SVM parameters for aircraft engine damage repair
Size of real data set (number of samples)/model | 5000 | 10,000 | 20,000 | |
|---|---|---|---|---|
Single diagnosis time (s) | ISMA-SVM | 42.3 ± 2.1 | 89.5 ± 3.2 | 185.2 ± 4.5 |
PSO-SVM | 58.6 ± 2.5 | 132.4 ± 3.8 | 310.7 ± 5.2 | |
p value | < 0.001 | < 0.001 | < 0.001 | |
Memory usage (GB) | ISMA-SVM | 3.8 ± 0.2 | 7.5 ± 0.3 | 14.8 ± 0.5 |
PSO-SVM | 4.2 ± 0.3 | 8.6 ± 0.4 | 17.9 ± 0.6 | |
p value | 0.003 | 0.002 | 0.001 | |
With the growth of data volume time(%) | ISMA-SVM | / | 111.6 | 106.9 |
PSO-SVM | / | 126.0 | 134.7 | |
p value | / | 0.005 | 0.003 | |
