Battery State of Charge Estimation Based on Regular/Recurrent Gaussian Process Regression

    •  Sahinoglu, G.O., Pajovic, M., Sahinoglu, Z., Wang, Y., Orlik, P.V., Wada, T., "Battery State of Charge Estimation Based on Regular/Recurrent Gaussian Process Regression", IEEE Transactions on Industrial Electronics, DOI: 10.1109/​TIE.2017.2764869, Vol. 65, No. 5, pp. 4311-4321, October 2017.
      BibTeX TR2017-124 PDF
      • @article{Ozcan2017oct,
      • author = {Sahinoglu, Gozde Ozcan and Pajovic, Milutin and Sahinoglu, Zafer and Wang, Yebin and Orlik, Philip V. and Wada, Toshihiro},
      • title = {Battery State of Charge Estimation Based on Regular/Recurrent Gaussian Process Regression},
      • journal = {IEEE Transactions on Industrial Electronics},
      • year = 2017,
      • volume = 65,
      • number = 5,
      • pages = {4311--4321},
      • month = oct,
      • doi = {10.1109/TIE.2017.2764869},
      • url = {}
      • }
  • MERL Contacts:
  • Research Areas:

    Artificial Intelligence, Electric Systems, Machine Learning, Signal Processing


This paper presents novel machine-learning based methods for estimating state of charge (SoC) of Lithium-ion (Li-ion) batteries which use Gaussian process regression (GPR) framework. The measured battery parameters, such as voltage, current, temperature, are used as inputs for regular GPR whereas the SoC estimate at the previous sample is fed back, and incorporated into the input vector for recurrent GPR. The proposed methods consist of two parts. In the first part, training is performed wherein the optimal hyperparameters of a chosen kernel function are determined to model data properties. In the second part, online SoC estimation is carried out according to the trained model. One of the practical advantages of a GPR framework is to quantify estimation uncertainty, and hence to enable reliability assessment of the battery SoC estimate. The performance of the proposed methods is evaluated by using simulated dataset and two experimental datasets, one with constant and the other with dynamic charge and discharge currents. The simulations and experimental results show the superiority of the proposed methods in comparison to state-of-the-art techniques including support vector machine (SVM), relevance vector machine (RVM) and Neural Network (NN).