Predictive Prognostics for Lithium-ion Battery for Electric Vehicle Applications

Planed funding: VEHICLE & HALFBACK projects co-financed by “INTERREG V Upper Rhine”. Thesis supervisors: Tedjani MESBAHI, Ahmed SAMET Contact: tedjani.mesbahi@insa-strasbourg.fr; ahmed.samet@insa-strasbourg.fr Workplace: INSA Strasbourg Laboratory: ICube - UMR 7357 Teams: Systèmes et Microsystèmes Hétérogènes (SMH), Science des Données et Connaissances (SDC) Starting date: September 1, 2019 VEHICLE project:

The VEHICLE project (Advanced Lithium-ion battery/supercapacitor Hybrid energy storage system with synchronous reluctance machine for electric vehicle applications) aims to develop adapted solutions to on-board energy storage systems through the hybridization of energy sources and the use of innovative machines for electric vehicles. The Upper Rhine region is home to leading-edge laboratories in the field of electric traction. VEHICLE is built to combine existing complementary expertise and create synergies to lead to the development of innovations, and establish a new research consortium in the Upper Rhine region. French, German and British scientists are involved with a network of 3 main academic partners and 6 associate partners. The VEHICLE project is developed as part of the INTERREG V Upper Rhine program and the Offensive Science initiative. It is co-financed in the context of this initiative by the Grand Est Region in France, the Baden-Württemberg and Rhineland-Palatinate in Germany.

HALFBACK project:

The main goal of the HALFBACK project is to assure highly available manufacturing processes, by forecasting failures of machines, tools, product quality loss, resource flow problems, etc. and by scheduling maintenance, component replacing, process re-planning, and even take over the production by another factory, in an optimized and intelligent way.

Data will be gathered using sensors located on the machines and tools. Additional information will be collected from various tools, the manufacturing environment, the product itself as well as the machine operator’s experience. Big data algorithms will use the collected data to understand the process and to learn from the experience of the operators and semantic technologies will be used to predict machine damage, quality loss or maintenance demands in the future. This will allow the company to act before the manufacturing process stops.

three academic partners involved in HALFBACK Project, along with several SMEs and one SME cluster. The Academic partners are INSA Strasbourg, University of Strasbourg and Hochschule Furtwangen University.

Context of the study:

The energy challenge is one of the main locks to the development of efficient, less polluting and economically viable means of transport with a rational use of the world's natural resources. In this context, vehicle manufacturers in the world are undergoing unprecedented technological change [1], [2]. In the transport field, the voices of progress are linked, among other things, to the hybridization and electrification of vehicles [3]. In both configurations, the effective operation of the systems is mainly linked to the availability of the on-board network and therefore of the battery. The energy storage systems used in the latest generation of hybrid and electric vehicles

are mainly based on Lithium-ion (Li- ion) technology. It is because it has a high specific energy (≈200Wh/kg) that this technology has established itself on the market of automotive [4].

Scientific objectives:

In the case of hybrid vehicles, and even more so in the case of all-electric powertrains, the on-board energy storage system remains the weak link: very expensive, limited in driving range, slow to recharge, main causes of the over-costs, ... The challenge for any car manufacturer wishing to develop a clean vehicle is therefore not only to optimize its electric powertrains, both in terms of cost and range, but also to bring the battery into line with the life of the vehicle [1],[3]. Battery lifetime is therefore a crucial element for the development of electric vehicles under acceptable cost conditions. In this context, the failure of battery could lead to serious inconvenience, performance deterioration, accelerated aging and costly maintenance [5],[6]. For that, the prognostics of on-board energy storage system aims to predict the remaining lifetime of a battery and to perform necessary maintenance service if necessary, using the past and current information. A reliable prognostic model should be able to accurately predict the future state of the battery such that the maintenance service could be scheduled in advance [7].

Figure 1: Proposed approach for battery ageing prognistic Mesbahi et al. [8] have developed a multi-physical model that integrates electrical, thermal and ageing aspects into a single algorithm, making it possible to better analyses the lifetime system by accurately simulating the behavior of Lithium-ion batteries. Based on the perspectives of this work, the proposed thesis aims to mix battery life extension and battery ageing analysis with new techniques of machine learning. Indeed, several techniques based on data mining and machine learning have been proposed under the HALFBACK project to predict the failure or loss of product quality of a machine based on sequential data [9]. Batteries and BMS are kinds of complex, sensor equipped machine that require data analysis techniques on sequential data as those already developed in Halfback project.

To the best of our Knowledge, only a few research studies have used Recurrent Neural Network (RNN) [10] approaches on this application despite being one of the most performing machine leaning techniques [11]. Therefore, we aim to use RNN type approaches like LSTM [10] to anticipate ageing and predict the loss of performances. For experimental side, these approaches should be deployed on microcontroller chip fixed on the battery.

Candidate's skills:

Specific knowledge: Programming skills in machine learning libraries such as tensorflow, keras to cite a few. Also, knowledge in CAN-BUS technology communicates, matlab, PSIM and labview. Basic knowledge of advanced automatic strategies and multi-physical modeling are very appreciated.

Desired education: Master or Engineer (Bac + 5) with a specialization in Electrical Engineering, Computer Science, or Electronics and Automatics Engineering.

Desired personal skills: High motivation for innovation and the search for operational solutions in an industrial context. Motivation for the combination of simulation and experimentation.

References:

[1] R. Sadoun, “Intérêt d’une Source d’Energie Electrique Hybride pour véhicule électrique urbain – dimensionnement et tests de cyclage,” Thèse de Doctorat, ECOLE CENTRALE DE LILLE, 2013.

[2] J. Zhang et al., “An Optimal Design and Analysis of a Hybrid Power Charging Station for Electric Vehicles Considering Uncertainties,” IECON 2018 - 44th Annu. Conf. IEEE Ind. Electron. Soc., vol. 1, pp. 5147–5152, 2019.

[3] D. V. Do, “Diagnostic de batteries Lithium ion dans des applications embarquées Table des matières,” Thèse de Doctorat, Université de Technologie de Compiègne, 2010.

[4] P. J. Kollmeyer, “Modeling of Low-Temperature Operation of a Hybrid Energy Storage System with a Butler-Volmer Equation Based Battery Model,” 2016.

[5] A. Downey, Y. H. Lui, C. Hu, S. Laflamme, and S. Hu, “Physics-based prognostics of lithium-ion battery using non-linear least squares with dynamic bounds,” Reliab. Eng. Syst. Saf., vol. 182, no. March 2018, pp. 1–12, 2019.

[6] L. Ren, L. Zhao, S. Hong, S. Zhao, H. Wang, and L. Zhang, “Remaining Useful Life Prediction for Lithium-Ion Battery: A Deep Learning Approach,” IEEE Access, vol. 6, pp. 50587–50598, 2018.

[7] H. Li, D. Pan, and C. L. P. Chen, “Intelligent prognostics for battery health monitoring using the mean entropy and relevance vector machine,” IEEE Trans. Syst. Man, Cybern. Syst., vol. 44, no. 7, pp. 851–862, 2014.

[8] T. Mesbahi, N. Rizoug, P. Bartholomëus, R. Sadoun, F. Khenfri, and P. Le Moigne, “Dynamic model of li-ion batteries incorporating electrothermal and ageing aspects for electric vehicle applications,” IEEE Trans. Ind. Electron., vol. 65, no. 2, pp. 1298–1305, 2018.

[9] C. Sellami, A. Samet, M-A. Bach Tobji:Frequent Chronicle Mining: Application on Predictive

Maintenance. ICMLA 2018: 1388-1393. [10] Ian J. Goodfellow, Yoshua Bengio, Aaron C. Courville: Deep Learning. Adaptive computation and machine

learning, MIT Press 2016, ISBN 978-0-262-03561-3, pp. 1-775 [11] Zhao, G., Zhang, G., Liu, Y., Zhang, B., & Hu, C. (2017, June). Lithium-ion battery remaining useful life

prediction with deep belief network and relevance vector machine. In 2017 IEEE International Conference on Prognostics and Health Management (ICPHM) (pp. 7-13). IEEE.