Deep Learning Neural Network Model for Tunnel Ground Surface Settlement Prediction Based on Sensor Data

Table 3

The results of different methods for predicting the settlement of various points on the surface of the tunnels for Ningbo Metro Line 3 (NB) or Zhuhai city tunnel (ZH).

Point number	Proposed			BPNN			LSTM			EMD-LSTM
	MAE	RMSE	MAPE%	MAE	RMSE	MAPE%	MAE	RMSE	MAPE%	MAE	RMSE	MAPE%
180 (ZH)	0.18	0.19	0.53	0.24	0.27	0.71	0.47	0.49	1.42	0.37	0.40	1.10
184 (ZH)	0.24	0.25	0.71	0.32	0.33	0.96	1.56	1.58	4.68	0.30	0.32	0.90
186 (ZH)	0.15	0.18	0.48	0.18	0.21	0.59	0.57	0.62	1.84	0.15	0.17	0.49
189 (ZH)	0.07	0.08	0.27	0.13	0.15	0.50	0.82	0.83	3.07	0.12	0.14	0.44
205 (ZH)	0.27	0.28	1.05	0.57	0.60	2.22	1.62	1.65	6.30	0.57	0.58	2.23
522 (NB)	0.36	0.44	2.29	0.47	0.63	2.86	1.54	1.64	9.41	0.49	0.58	2.99
554 (NB)	0.20	0.27	1.37	0.35	0.45	2.36	0.55	0.68	3.67	0.43	0.50	2.92
559 (NB)	0.20	0.25	2.42	0.45	0.51	5.25	0.48	0.56	5.64	0.23	0.27	2.66
569 (NB)	0.49	0.62	2.14	0.76	0.87	3.03	0.65	0.85	2.52	1.07	1.11	4.33
570 (NB)	0.30	0.35	1.84	0.47	0.63	2.86	1.04	1.21	6.28	0.33	0.38	2.04