Fig. 1

“×”configuration

Fig. 2

Schematic diagram of principle of integrated navigation system

Fig. 3

Architecture of DLinear

Fig. 4

Informer model architecture

Fig. 5

Training flowchart of beam predictor

Fig. 6

Prediction flowchart of the beam predictor

Fig. 7

ImuFOG/GPS integrated navigation trajectory

Fig. 8

DVL beam velocity

Table 1

Parameters and values of the DLinear-Informer model

Fig. 9

Variation of RMSE of DVL beam predictions with the number of beam measurements of past DVL

Fig. 10

Beam4 outputs of DVL, MLP, CNN, LSTM, and DLinear-Informer models

Fig. 11

MAE and STD of MLP, CNN, LSTM, and DLinear-Informer models in the case of a missing beam

Table 2

Specific values of MAE and STD for four models in the absence of a beam

Fig. 12

Absolute position error of a missing beam

Fig. 13

Beam3 and Beam4 output of DVL, MLP, CNN, LSTM, and DLinear-Informer models

Fig. 14

MAE and STD of MLP, CNN, LSTM, and DLinear-Informer models in the case of two missing beams

Table 3

MAE and STD for four models in absence of two beams

Fig. 15

Absolute position error of two missing beams

Fig. 16

Beam2~4 output of DVL, MLP, CNN, LSTM, and DLinear-Informer models

Fig. 17

MAE and STD of MLP, CNN, LSTM, and DLinear-Informer models in the case of three missing beams

Table 4

MAE and STD for 4 models in absence of three beams

Fig. 18

Absolute position error of three missing beams

Table 5

DLinear-Informer p-values and t-values for t-test analyses in three beam missing scenarios