Extended Physics-Informed Neural Networks for Solving Parameterized Cyclic Voltammetry

Cyclic voltammetry is a widely used technique in electrochemistry. In recent years, Physics-informed Neural Networks (PINNs), combining data-driven and physical model constraints, have been applied to solve cyclic voltammetry models in electrochemical systems. However, the network non-reusability of PINNs greatly limits the value and efficiency of PINNs in simulating cyclic voltammetry. This paper proposes an extended method for PINNs, which parameterizes the thin-layer length factor in cyclic voltammetry and incorporates it into PINNs training. The method enables the network model to predict the corresponding partial differential equation solutions under any thin layer length factor, significantly improving computational efficiency. The model’s reliability was verified by comparing the predicted cyclic voltammetry results with the new approach with high-precision calculation results.