Table 4 The predictive results of various models for the lithium-ion battery.

From: Application of state of health estimation and remaining useful life prediction for lithium-ion batteries based on AT-CNN-BiLSTM

Method

Battery

RULae

Time(s)1

MAPE

RMSE

R2

LSTM

B5

1

16.2655

0.6307

0.0069

0.9947

B6

2

14.9719

1.0187

0.0122

0.9901

B7

/

14.9889

0.4667

0.0058

0.9946

BiLSTM

B5

1

25.8738

0.5288

0.0062

0.9956

B6

1

19.4240

0.8826

0.0112

0.9917

B7

/

19.5776

0.3968

0.0054

0.9954

CNN-LSTM

B5

2

22.9955

0.6235

0.0064

0.9954

B6

2

18.6910

1.1120

0.1169

0.9909

B7

/

18.9583

0.4906

0.0056

0.9950

CNN-BiLSTM

B5

1

27.8186

0.5163

0.0062

0.9956

B6

0

24.7927

0.7981

0.0110

0.9919

B7

/

24.7109

0.4076

0.0054

0.9954

CNN-LSTM-Attention

B5

1

26.2932

0.5901

0.0065

0.9952

B6

1

22.4589

0.9003

0.0116

0.9910

B7

/

22.0530

0.4646

0.0057

0.9949

CNN-BiLSTM-Attention

B5

0

30.7099

0.5106

0.0060

0.9959

B6

0

30.2102

0.8389

0.0109

0.9921

B7

/

30.4226

0.4018

0.0053

0.9956

  1. 1Time represents the period taken to train the prediction model.