|
Physics-Based Battery Insight
Artiphy is developing next-generation modelling technology for lithium-ion batteries.
Our work combines advances in electrochemical modelling and scientific machine learning to create ultra-fast neural surrogates of battery physics.
These models enable new capabilities across battery management, design and analytics by making high-fidelity physics models accessible in real-time applications.
We are a team based at the University of Portsmouth with world-class expertise in electrochemical modelling and scientific machine learning
Professor of Industrial & Applied Mathematics
Research focuses on electrochemical battery modelling and scientific computing.
Professor of Probability & Statistical Physics
Expertise in applied mathematics, complex systems and data-driven modelling.
Research Fellow
Research spans machine learning, statistical physics and scientific modelling.
PhD Researcher
Working on surrogate modelling techniques for electrochemical battery systems.
Artiphy may be able to help you if you need to:
Electrochemical battery models provide deep insight into the internal behaviour of lithium-ion cells. They describe ion transport, reaction kinetics and electrode dynamics using systems of coupled differential equations. These models are powerful but computationally demanding, which limits their use in real-time applications. Artiphy addresses this limitation by constructing neural surrogates of electrochemical battery models. These surrogates replicate the behaviour of the underlying physics model while running dramatically faster, making high-fidelity modelling accessible in applications where traditional solvers are impractical.
Artiphy models are neural surrogates designed to reproduce the behaviour of electrochemical battery models with extremely high computational efficiency. Once constructed, these surrogate models can evaluate battery behaviour rapidly without solving the underlying system of equations. This enables physics-based prediction and analysis to be performed orders of magnitude faster than with classical modelling approaches. The resulting models combine the interpretability of physics-based modelling with the speed and efficiency required for real-time applications.
Artiphy models retain access to physically meaningful internal variables such as lithium concentration and reaction rates.
Model sizes are typically under 100 KB, enabling deployment on embedded processors and battery management hardware.
Models accept arbitrary current profiles and operating conditions.
Fast evaluation enables advanced statistical techniques such as Bayesian parameter inference and uncertainty quantification.