Organisational framework in the ENERGETIC project
Review available battery technologies and select the most suitable to use as a case study in this project, considering first and second life applications; Propose advanced battery sizing methodology including a wider range of considerations than current methods; Evaluate the feasibility of using low cost (<5€) ultrasonic transducers rather than already-demonstrated more expensive (~50€) ones and characterize battery cells using these; Thermally characterise battery cells using in-situ temperature sensors; Train AI computer vision algorithms using synthetic thermal images to detect anomalies; Collection and processing of experimental data from the above objectives which may be used in WP2 and WP3 to evaluate the extent to which the BMS may be augmented by this information
Identify the physical parameter (indicators), develop advanced physical models, develop new AI models for battery SoX ageing, develop semi-supervised approach for battery degradation; Assess the SoX and RUL of a Lithium battery continuously, one of the best approaches seems to be a digital twin, were physical parameters or indicators are used in advanced physical models, which, in combination with AI methods, will be able to develop a semi-supervised approach. The battery degradation is influenced by multiple parameters thus requiring à multiphysics model including a thermal modelling. An adapted battery cooling systems seems to have an important influence on the battery lifetime; Moreover, as multiple battery cells are combined to form modules and battery packs in which the cells have encounter slightly different electrical, chemical, and thermal conditions, it seems important to model the system in its totality, imposing a huge number of parameters to identify.
Develop new supervised machine learning model for battery SoX prediction using usage sensors time series data; Develop an unsupervised machine learning model for cell anomaly detection using thermal images; Explain using posthoc XAI previous models to identify SoX degradation patterns; Multimodal learning using both early and late information fusion to make use of previous models for a robust SoX prediction.
Design an innovative smart BMS integrating performance results from a novel digital-twin based on an ensemble of hybrid explainable AI and expert models. A battery hardware solution will be proposed that is compatible with the proposed BMS, applying operational and thermal policies for better heat monitoring and extended battery life.
To connect the BMSto the internet (security, 5G, blockchain), Edge BMS, predictive maintenance for battery using Cloud.
To validate the algorithms developed in the project in use cases representing actual operating conditions at which the algorithms should perform with satisfying performances (computation time, accuracy, battery operation safety…) under 1st life (mobility) and 2nd life (stationary) applications. KPIs will be defined to assess performance and cost acceptability for using the algorithms in 1st and 2nd life applications.
Increase the visibility and support the impact of ENERGETIC and its results; Design, manage and deliver the communication and dissemination of the project; Maximize the project’s results by disseminating and communicating updates and the acquired knowledge to relevant identified stakeholders and in terms that are readily understandable to stakeholdersin industry, research and academia, policymakers and wider public in order to accelerate the implementation of the research findings.
The aim of this WP is to adequately coordinate and organize the project at strategic level, ensuring project follow-up (project progress control and planning); continuous update of the project status to the commission officers; decision making procedures and suitable project administration; and assure data management, protection and privacy during all the project duration.