Enhanced physics-constrained deep neural networks for modeling vanadium redox flow battery

Numerical simulation has become indispensable in advancing cost-effective process optimization and control of flow batteries. We propose an enhanced version the physics-constrained deep neural network (PCDNN) approach (He et al., 2022) to provide high-accuracy voltage predictions vanadium redox batteries (VRFBs). The purpose PCDNN is enforce physics-based zero-dimensional (0D) VRFB model a assure generalization for various battery operation conditions. However, limited by simplifications 0D model, cannot capture sharp changes extreme SOC regions. To improve accuracy prediction at ranges, we introduce second (enhanced) DNN mitigate errors carried from itself call resulting (ePCDNN). By comparing with experimental data, demonstrate that ePCDNN can accurately response throughout charge–discharge cycle, including tail region discharge curve. loss function training designed be flexible adjusting weights DNN. This allows framework transferable systems variable physical fidelity.