Table 10 Adding relative position encoding model performance effect.

From: Deep multiscale feature fusion network with dual attention for rolling bearing remaining useful life prediction

Method

Bearing2_2

Bearing2_5

RMSE

MAE

R2

RMSE

MAE

R2

Transformer-encoder

0.07556

0.06476

0.92801

0.08883

0.06501

0.90374

SE-transformer-encoder

0.07506

0.06257

0.92489

0.09520

0.07226

0.88870

GCU-transformer

0.09975

0.08664

0.87783

0.11141

0.08506

0.84853

TCN-MA

0.09868

0.08291

0.91576

0.16079

0.13466

0.68240

The paper (not position)

0.07924

0.06578

0.91421

0.08689

0.06923

0.90823

The paper

0.05301

0.04159

0.96186

0.07505

0.06182

0.93149

  1. Significance values are bold.
Back to article page