کنفرانس مهندسی برق ایران

صفحه اصلی / سی و سومین کنفرانس بین المللی مهندسی برق

Temperature Prediction of Lithium-Ion Batteries for Thermal Management Systems Using Graph Convolutional Networks

نویسندگان :

Sepehr Ghalebi¹ Elaheh Sadat Ahmadi Mousavi² Farzaneh Abdollahi³ Farschad Torabi⁴

1- Amirkabir University of Technology (Tehran Polytechnic) 2- Amirkabir University of Technology (Tehran Polytechnic) 3- Amirkabir University of Technology (Tehran Polytechnic) 4- K.N.Toosi University of Technology

کلمات کلیدی :

Lithium-ion Battery،Battery Management System،BMS،Battery Thermal Management System،BTMS،Graph Neural Network،Graph Convolutional Networks،Deep Learning

چکیده :

In this paper, an innovative approach for predicting the temperature of lithium-ion batteries (LIBs) for thermal management applications is presented utilizing graph neural networks (GNNs). This study evaluates the employment of graph convolutional networks (GCNs) for predicting temperature trends and fluctuations across several cycles and compares its performance with well-established methods, such as long short-term memory (LSTM) and convolutional neural networks (CNN). By assigning each sample to a node and modeling relationships between consecutive samples with graph edges, the GCN effectively captures temporal patterns in the data. The presented model architecture includes graph convolutional layers, dropout, and a fully connected output layer. The GCN was evaluated using different training-validation splits, and the results indicated that the proposed approach achieves higher accuracy compared to LSTM and CNN, with the RMSE of lower than 1.6%. The results indicate the proposed GCN computational efficiency and precision, making it a reliable method in thermal management applications.