Table 7 Comparison of battery SOH prediction performance based on LOOCV.

From: Estimation of lithium-ion battery health state using MHATTCN network with multi-health indicators inputs

 

Model

Evaluation metrics

 

MAE

MSE

\(\overline{R^{2}}\)

RMSE

MAPE

Dataset 1

MHAT-TCN

0.0064

0.0001

0.9968

0.0118

1.0430

\({\textrm{TCN}}\)

0.0082

0.0002

0.9958

0.0133

1.3049

GRU

0.0146

0.0003

0.9921

0.0184

2.1709

LSTM

0.0168

0.0004

0.9902

0.0205

2.5276

Dataset 2

MHAT-TCN

0.0107

0.0003

0.9954

0.0175

2.3723

\({\textrm{TCN}}\)

0.0203

0.0007

0.9891

0.0270

3.5583

GRU

0.0196

0.0008

0.9889

0.0273

4.2291

LSTM

0.0206

0.0009

0.9875

0.0289

4.4806

Dataset 3

MHAT-TCN

0.0073

0.0002

0.9960

0.0133

1.2171

\({\textrm{TCN}}\)

0.0112

0.0006

0.9869

0.0238

1.8182

GRU

0.0138

0.0005

0.9894

0.0213

2.5824

LSTM

0.0147

0.0005

0.9892

0.0215

2.6284

Dataset 4

MHAT-TCN

0.0118

0.0003

0.9930

0.0171

1.6818

\({\textrm{TCN}}\)

0.0110

0.0006

0.9844

0.0254

2.1362

GRU

0.0128

0.0005

0.9896

0.0207

2.2425

LSTM

0.0137

0.0005

0.9878

0.0224

2.4834

Back to article page