Scholarlink

<p>Electric Vehicles (EVs) rely heavily on battery systems, making accurate State-of-Health (SoH) prediction essential for ensuring performance, reliability, and safety. However, prediction accuracy can be affected by varying operational and environmental conditions. To address this, a robust framework is suggested using a Quantum Spatial Graph Convolutional Neural Network (QSGCNN) optimized by the Dollmaker Optimization Algorithm (DmOA). Data from the Centre for Advanced Life Cycle Engineering (CALCE) is first normalized and structured using Neural Causal Graph Collaborative Filtering (NCGCF). QSGCNN captures spatial-temporal dependencies in the battery data for SoH estimation, while DmOA fine-tunes its parameters for improved prediction performance. The suggested QSGCNN-DmOA model is implemented in MATLAB and evaluated against Gated Recurrent Unit (GRU), Deep Neural Network-Long Short-Term Memory (DNN-LSTM), Emperor Penguin-Bidirectional LSTM (EP-BLSTM), Bidirectional Recurrent Neural Network-LSTM (biRNN-LSTM), and Convolutional Neural Network-LSTM (CNN-LSTM). Simulation outcomes demonstrate that the suggested approach achieves superior accuracy, with Root Mean Squared Error (RMSE) of 0.17, Mean Absolute Error (MAE) of 0.14, and Mean Absolute Percentage Error (MAPE) of 3.9, confirming its effectiveness and reliability for SoH prediction in EV batteries.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

State-of-Health Prediction for Electric Vehicle Battery Systems with Quantum Spatial Graph Convolutional Neural Network Optimized Using Dollmaker Optimization Algorithm

  • 期刊论文

  • 发表日期：

    2025年10月30日

查看全文
  • V. Rengarajan,
  • T. Anuradha

摘要

Electric Vehicles (EVs) rely heavily on battery systems, making accurate State-of-Health (SoH) prediction essential for ensuring performance, reliability, and safety. However, prediction accuracy can be affected by varying operational and environmental conditions. To address this, a robust framework is suggested using a Quantum Spatial Graph Convolutional Neural Network (QSGCNN) optimized by the Dollmaker Optimization Algorithm (DmOA). Data from the Centre for Advanced Life Cycle Engineering (CALCE) is first normalized and structured using Neural Causal Graph Collaborative Filtering (NCGCF). QSGCNN captures spatial-temporal dependencies in the battery data for SoH estimation, while DmOA fine-tunes its parameters for improved prediction performance. The suggested QSGCNN-DmOA model is implemented in MATLAB and evaluated against Gated Recurrent Unit (GRU), Deep Neural Network-Long Short-Term Memory (DNN-LSTM), Emperor Penguin-Bidirectional LSTM (EP-BLSTM), Bidirectional Recurrent Neural Network-LSTM (biRNN-LSTM), and Convolutional Neural Network-LSTM (CNN-LSTM). Simulation outcomes demonstrate that the suggested approach achieves superior accuracy, with Root Mean Squared Error (RMSE) of 0.17, Mean Absolute Error (MAE) of 0.14, and Mean Absolute Percentage Error (MAPE) of 3.9, confirming its effectiveness and reliability for SoH prediction in EV batteries.