Analysis of Diaphragm Wall Deflection Induced by Excavation Based on Machine Learning

Table 5

MSEs of different algorithms on 10-fold cross validation (unit: mm²).

Subset no.	Task 1			Task 3
	BPNN	LSTM	GRU	BPNN	LSTM	GRU
Short-term prediction tasks
1	3.47	2.21	0.91	1.39	1.22	0.38
2	12.49	2.36	1.22	1.86	2.07	0.64
3	7.14	3.79	0.78	10.46	6.70	0.39
4	4.11	1.24	0.53	0.18	0.38	0.30
5	9.22	1.86	0.61	0.65	0.76	0.37
6	5.41	1.35	0.46	32.95	20.14	0.73
7	17.25	8.56	1.76	0.27	0.27	1.56
8	125.36	4.86	1.07	0.41	0.19	0.69
9	624.14	5.93	1.27	13.84	5.00	1.62
10	63.04	10.09	2.50	0.27	0.26	1.46
Avg.	87.16	4.23	1.11	6.23	3.70	0.81
Subset no.	Task 2			Task 4
	BPNN	LSTM	GRU	BPNN	LSTM	GRU
Long-term prediction tasks
1	181.38	19.47	3.50	8.29	5.51	2.21
2	44.18	14.67	6.99	6.61	4.85	2.36
3	75.09	14.38	6.94	7.78	6.34	3.79
4	44.07	4.42	2.78	5.01	5.05	1.24
5	101.18	2.68	3.53	4.67	3.44	1.86
6	7.04	41.55	2.69	8.05	18.82	1.35
7	402.82	2.23	5.55	16.89	6.60	8.56
8	185.21	3.67	9.34	17.76	1.34	4.86
9	98.77	24.87	9.24	29.57	12.02	5.93
10	194.29	4.17	6.62	23.31	3.95	10.09
Avg.	133.40	13.21	5.72	12.79	6.79	4.23