Estimation of Vehicle Dynamic Response from Track Irregularity Using Deep Learning Techniques

Table 1

Performance of different models.

Model	RMSE	MAE	ρ	TIC	params	FLOPs (MB)	Time (s)
LSTM	1361.8107	933.4527	0.7492	0.1129	119.054 KB	57.876	2.23
CNN-LSTM	1177.3689	818.7498	0.8062	0.1031	364.0 K	178.381	5.67
CA-CNN-LSTM	1116.7266	762.0530	0.8287	0.1023	364.058 K	179.155	5.95
CNN-MUSE	1062.8726	737.1695	0.8418	0.0959	1.222 M	613.581	4.67
CA-CNN-MUSE	1019.6666	705.9139	0.8531	0.0925	1.226 M	615.993	4.71