You are on page 1of 2

Data Visualization for Digital Twin-based BMS

using Unreal Engine


Robin Matthew Medina ∗ , Judith Nkechinyere Njoku ∗ , Paul Michael Custodio ∗ , Cosmas Ifeanyi Nwakanma†
Jae-Min Lee ∗ , and Dong-Seong Kim ∗
∗ IT Convergence Engineering, † ICT Convergence Research Center, Kumoh National Institute of Technology, Korea.

Abstract—This paper proposes to visualize a Battery Model a GRU-RNN model is employed. The GRU-RNN model
for Battery Management Systems (BMS) using Unreal Engine comprises a single GRU layer and two Dense layers. The
for Digital Twin Technology. By utilizing an interactive 3D model is trained using the battery dataset, aiming to achieve
platform, the proposed approach allows for real-time monitoring
and prediction of the battery performance data, enabling easy high prediction accuracy for SOC.
identification of inefficiencies. The visualization is achieved using
only static input as an alternative for the SOC prediction using B. Integration with Unreal Engine
Data-Driven AI Model. Once the AI model is trained and exhibits satisfactory accu-
Index Terms—Digital Twin, 3D Model, Battery Management racy, it is integrated into the Unreal Engine platform. Unreal
System
Engine offers advanced capabilities for creating immersive
I. I NTRODUCTION visual experiences. The prediction results obtained from the
trained AI model are used as input for the data visualization
As battery-powered technologies gain widespread adoption,
within Unreal Engine.
the optimization of battery performance and longevity be-
comes paramount [1]. Battery management systems (BMS) C. Creation of Battery 3D Model
play a crucial role in monitoring and controlling battery
As part of the visualization process, a simplified 3D model
performance, ensuring optimal usage and extended lifespan
of the battery is generated using 3DS Max. This model serves
[2].
as the main object for visualization and is imported into the
Building on the concept of Digital Twin, which provides a
Unreal Engine environment.
virtual dynamic model of a system, process, or service, we
propose harnessing the power of data visualization to extract
TABLE I
meaningful insights from battery-related information [3]. By BATTERY T HRESHOLD I DENTIFIER
utilizing data visualization techniques, battery manufacturers,
researchers, and engineers can make informed decisions to SOC Color
optimize battery performance [4]. To achieve the requisite data Empty Battery (0%) Gray
Low Battery (1% - 25%) Red
for visualization, we develop a predictive model for battery Battery SOC (26% - 50%) Orange
parameters, focusing on the critical aspect of State-of-Charge Battery SOC (51% - 75%) Yellow
Battery SOC (76% - 100%) Green
estimation. In this regard, deep neural networks (DNN) have
demonstrated success in battery state estimation [5].
This paper introduces a novel approach utilizing the robust D. Visualization Implementation
capabilities of Unreal Engine, a powerful game engine known
Given the emphasis of this paper on the visualization aspect,
for its advanced 3D visualization features, to develop an im-
a static SOC input is utilized, representing a predefined range
mersive and interactive data visualization platform for digital
of SOC values. This input is modified based on the predicted
twin-based BMS. By leveraging Unreal Engine’s cutting-edge
SOC obtained from the AI model. The color of the battery also
technology, we aim to enhance the visualization experience
changes in line with a predefined threshold, as illustrated in
and provide a dynamic environment for analyzing and inter-
Table I. The Unreal Engine platform is leveraged to visualize
preting BMS data.
the battery model and the changing SOC input, providing an
II. M ETHODOLOGY interactive and immersive data visualization experience for
This section outlines the proposed methodology for devel- battery performance analysis.
oping and integrating the Digital Twin-based Battery Manage-
III. D ISCUSSION
ment System (BMS) visualization platform. The process flow
is illustrated in Fig 1. This section discusses the early results obtained in this
paper. These results demonstrate the significance of battery vi-
A. Dataset Selection and Model Training sualization in facilitating the identification of State-of-Charge
The battery dataset chosen for model training is the Pana- (SOC) through visual cues. By creating a basic 3D battery
sonic 18650. To predict the battery’s State-of-Charge (SOC), model using 3DS Max and importing it into Unreal Engine,
Fig. 1. Proposed Model

Fig. 2. Data Visualization of Battery in Unreal Engine

we established the foundation for battery visualization within ACKNOWLEDGMENTS


the digital twin-based BMS. This research was supported by Priority Research
As depicted in Fig. 2, the battery model undergoes color Centers Program through the NRF funded by the MEST
changes to represent different SOC levels. Currently, the color (2018R1A6A1A03024003), NRF-2022R1I1A1A010701058
changes are based on a static input defined by the user. and MSIT, Korea, under the Grand Information Technology
However, the ultimate goal is to incorporate real-time changes Research Center support program(IITP-2023-2020-0-01612)
in the battery model based on the predictions generated by the supervised by the IITP.
AI model.
The threshold range employed for visual data in Fig. 2 R EFERENCES
is derived from Table I. This visualization is a preliminary [1] Z. Cui, L. Wang, Q. Li, and K. Wang, “A comprehensive review
on the state of charge estimation for lithium-ion battery based
example, and further enhancements will be implemented to on neural network,” International Journal of Energy Research,
ensure precision, clarity, and effectiveness in presenting battery vol. 46, no. 5, pp. 5423–5440, 2022. [Online]. Available: https:
performance data. //onlinelibrary.wiley.com/doi/abs/10.1002/er.7545
[2] W. Wang, J. Wang, J. Tian, J. Lu, and R. Xiong, “Application of
digital twin in smart battery management systems,” Chinese Journal of
IV. C ONCLUSIONS Mechanical Engineering, vol. 34, no. 1, p. 57, Jun. 2021.
[3] S. Singh, M. Weeber, and K. P. Birke, “Implementation of battery digital
twin: Approach, functionalities and benefits,” Batteries, vol. 7, no. 4,
This study presented the visualization of the battery state of 2021. [Online]. Available: https://www.mdpi.com/2313-0105/7/4/78
charge in a virtual platform, by leveraging on digital twin tech- [4] T. Azfar, J. Weidner, A. Raheem, R. Ke, and R. L. Cheu, “Efficient pro-
nology. Future work will focus on refining the visualization cedure of building university campus models for digital twin simulation,”
IEEE Journal of Radio Frequency Identification, vol. 6, pp. 769–773,
aspects, such as incorporat ing additional features, improving 2022.
the precision of visual data, and enhancing the overall ef- [5] J. N. Njoku, R. M. Medina, N. E. Chinaechetam, C. I. Nwakanma, J.-
fectiveness of the visualization platform. These advancements M. Lee, and D.-S. Kim, “Analysis of deep neural networks-based digital
twin for lithium-ion batteries,” in Korean Institute of Communications and
will ensure that the proposed method provides an efficient and Information Sciences (KICS Fall 2022) Conference, 2022, pp. 665–666.
intuitive means for battery analysis and decision-making.

You might also like