Sign in Subscribe
Batteries

Advances in Battery Research Yield Improved State Estimation for Energy Storage Power Stations

Strategic Insights

— 2 min read

Researchers from Shenyang University of Technology have made significant progress in developing an improved Transformer structure for estimating the state of charge (SOC) of lithium batteries in energy storage power stations. The study, funded by the National Natural Science Foundation of China and the Science Research Project of Hebei Education Department, employed a Time Delay Second Estimation (TDSE) algorithm to optimize the improved Transformer model. The team's innovative particle filter algorithms were designed to handle nonlinearity, uncertainty, and dynamic changes in predicting remaining battery life. The results showed that the model's maximum SOC estimation error was 2.68% at 10°C and 2.15% at 30°C for LiNiMnCoO positive electrode datasets. For LiFePO positive electrode datasets, the maximum error was 2.79% at 10°C (average 1.25%) and 2.35% at 30°C (average 0.94%).

Key Takeaways:

  • The improved Transformer structure was employed to estimate the battery's state of charge (SOC) in energy storage power stations.
  • The Time Delay Second Estimation (TDSE) algorithm optimized the improved Transformer model to overcome traditional models' limitations in extracting long-term dependency.
  • Innovative particle filter algorithms were proposed to handle nonlinearity, uncertainty, and dynamic changes in predicting remaining battery life.
  • The model's maximum SOC estimation error was 2.68% at 10°C and 2.15% at 30°C for LiNiMnCoO positive electrode datasets.
  • For LiFePO positive electrode datasets, the maximum error was 2.79% at 10°C (average 1.25%) and 2.35% at 30°C (average 0.94%).
  • The particle filter algorithm predicted battery capacity with 98.34% accuracy and an RMSE of 0.82%.
  • The improved model enables advanced battery state prediction, enhancing the adaptability and robustness of lithium battery state analysis.

Statistics:

  • Maximum SOC estimation error: 2.68% at 10°C, 2.15% at 30°C for LiNiMnCoO positive electrode datasets.
  • Maximum SOC estimation error: 2.79% at 10°C (average 1.25%), 2.35% at 30°C (average 0.94%) for LiFePO positive electrode datasets.
  • Predicted battery capacity accuracy: 98.34%.
  • Root Mean Square Error (RMSE): 0.82%.
  • LiNiMnCoO positive electrode datasets: 2.68% error at 10°C, 2.15% error at 30°C.
  • LiFePO positive electrode datasets: 2.79% error at 10°C, 2.35% error at 30°C.

Sources:

  • Performance Analysis of Battery State Prediction Based on Improved Transformer and Time Delay Second Estimation Algorithm. Batteries, 2025,11(7):262. (Batteries - http://www.mdpi.com/journal/batteries).
  • NewsRx. Shenyang University of Technology Researchers Detail Research in Battery Research (Performance Analysis of Battery State Prediction Based on Improved Transformer and Time Delay Second Estimation Algorithm). Energy Weekly News. August 15, 2025; p 484.

Read more