Research

Topics;

• Optimize EV charge and discharge scheduling to miximize user profit and convenience.

• Evaluate the impact of massive EV feet charging in the distribution networks.

• Probabilistic forecast for EV charging demand in local area based on machine learning.

• Reinforcement learning for battery charg and discharge

Related publication;

1. D. Kodaira, W. Jung, and S. Han, “Optimal Energy Storage System Operation for Peak Reduction in a Distribution Network Using a Prediction Interval,” IEEE Trans. Smart Grid, vol. 11, no. 3, pp. 2208–2217, 2020. link

2. D. Kodaira and J. Kondoh, "Probabilistic Forecasting Model for Non-normally Distributed EV Charging Demand," 2020 International Conference on Smart Grids and Energy Systems (SGES), 2020, pp. 623-626. link

3. M. Seo, D. Kodaira and S. Han, "Reduction of Computational Complexity for Optimal Electric Vehicle Schedulings," 2020 IEEE Power & Energy Society General Meeting (PESGM), 2020, pp. 1-5. Selected as one of the best papers! (top 5% in 2200 papers) link

4. M. A. Acquah, D. Kodaira, and S. Han, “Real-time demand side management algorithm using stochastic optimization,” Energies, vol. 11, no. 5, p.1166－1179, 2018. link

Topics;

• Develop prototype of the P2P trade system

Related publication;

Nothing yet

Project goal

The purpose of this project is to establish the theory of degradation function model and to build a prototype of data platform for creating degradation function model by machine learning, which evaluates the degree of degradation of lithium-ion batteries for electric vehicles (EVs).

Project description

Lithium-ion batteries are now commonly used in electric vehicles (EVs), and as EVs become more widespread in the future, not only their low environmental impact but also their economic benefits to users are becoming more important. For example, if the Leaf sold by Nissan is recharged every day, the life of the battery will be over in about three years and the battery will have to be replaced at the user's expense. The degree of degradation of lithium-ion batteries varies depending on various parameters, such as the temperature at which the battery operates, the amount of charge remaining, and the amount of current used during charging and discharging, making it difficult to make an accurate assessment in general. The degradation evaluation data is provided from battery manufacturers is generally obtained by repeating full charge and zero charge at a certain pace at a certain temperature, disassembling the battery, and measuring the voltage of the cells. The provided data from the manufactures are obtained in rigid experiment conditions, which can be evaluated only in laboratory. Therefore, the importance of the technology to diagnose the degree of degradation of lithium-ion battery of EV using only the parameters that can be measured during actual driving or power supply (operating temperature, data of battery terminal voltage, transition of remaining charge, etc.) is increasing. Dr. Kodaira has been working with Prof. Han Sekyung in Kyungpook National University in Korea on the degradation evaluation of lithium-ion batteries.

Prof. Han Sekyung has published a number of papers that have been cited more than 1,000 times and is regarded as an expert in the field. With the cooperation of Prof. Han, Dr. Kodaira developed a basic degradation function to evaluate the degree of degradation [1]. However, the current degradation function is effective only for data measured under certain conditions, and the degree of degradation cannot be evaluated correctly depending on the operating temperature and the accuracy of the measurement data, and a generalized theory that enables degradation evaluation from any measurement data has not yet been developed. Therefore, in this project, a generalized model to diagnose the degree of degradation of the lithium-ion battery of an EV by a machine learning model will be developed using only the data obtained from the actual operation of the EV. The training data for the machine learning model needs to be collected for a wide range of situational settings (e.g., cold and hot ambient temperature) for EVs in operation. Furthermore, the collected data has various formats and measurement accuracies depending on the vehicle, sensor type. Therefore, there is a need to construct a data platform that collects operational data on a large scale, converts it into an appropriate format that can be used for machine learning, stores it, and further operates it. This project aims to implement the basic architecture of the data platform, and to evaluate the battery degradation by the model learned through the developed data platform.

Related publication;

[1] D. Kodaira and S. Han, “Battery Degradation Platform and Model for Realistic Battery Use Cases,” in 2020 4th International Conference on Smart Grid and Smart Cities (ICSGSC), Aug. 2020, pp. 14–17.

[2] S. Han, S. Han, and H. Aki, “A practical battery wear model for electric vehicle charging applications,” Appl. Energy, vol. 113, pp. 1100–1108, Jan. 2014.

[3] S. Han and S. Han, “Economic feasibility of V2G frequency regulation in consideration of battery wear,” Energies, vol. 6, no. 2, pp. 748–765, 2013.

Topics;

• Indentify the distribution network topology using smart meter data

• Estimate the impedance in the lowe voltage distribution network

• Voltage control by reactive power

• Compensate the missing smart meter data

• Alternative power flow calculation by neural network

Related publication;

1. J. Park, D. Kodaira, K. A. Agyeman, T. Jyung, and S. Han, “Adaptive Power Flow Prediction Based on Machine Learning,” Energies, vol. 14, no. 13, p. 3842, Jun. 2021. link

2. D. Kodaira and S. Han, “Topology-based estimation of missing smart meter readings,” Energies, vol. 11, no. 1, p.11－28, 2018. link

3. S. Han, D. Kodaira*, S. Han, B. Kwon, Y. Hasegawa, and H. Aki, “An Automated Impedance Estimation Method in Low-Voltage Distribution Network for Coordinated Voltage Regulation,” IEEE Trans. Smart Grid, vol. 7, no. 2, pp. 1012–1020, 2016. (*Corresponding author) link