Battery State-of-Health and Remaining-Useful-Life Prediction from NASA Ageing Data

Li-ion cells lose capacity as they age. An electric-vehicle or energy-storage operator wants to know the state of health (SOH) of every cell now and its remaining useful life (RUL): the number of cycles until the capacity reaches the end-of-life limit (70 % of the rated 2 Ah = 1.4 Ah, the NASA criterion). This project uses the NASA PCoE Battery Data Set (18650 cells cycled to end of life at 24 degC) to estimate SOH with Gaussian-process regression from charge / discharge health indicators and to predict RUL with four methods: the classical double-exponential capacity-fade model fitted to the early cycles, a particle filter, a Gaussian-process trend and an LSTM capacity-sequence ensemble, all evaluated leave-one-battery-out with 90 % prediction intervals. It explains why cell B0018 (capacity "regeneration" after rest periods) is hard. A PyBaMM SEI-growth ageing simulation (run), a GNU Octave cross-check (run), a complete MATLAB version (written, not run) and a web application with five databases are included. Everything runs on a student laptop CPU in about 10 minutes plus a 210 MB download.

Category: Electrical and Electronics Engineering (Electric Vehicles, Battery Prognostics and Health Management, Gaussian Processes, Particle Filtering, Deep Learning, PyBaMM, MATLAB, GNU Octave, Python)

Electrical & Electronics Engineering Project EE05 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.