Physics Informed Bellman Neural Network for Energy Management Strategy of Hybrid Electric Vehicle

Due to lower carbon emission and free of range anxiety, hybrid electric vehicles (HEV) have been more and more popular. In HEV, energy management system (EMS) plays an vital role in managing the usage of fuel and electricity. Dynamic programming (DP) is one of the most well-known methods, which can obtain the optimal solution. However, the ‘curse of dimensionality‘ makes it difficult and challenging to implement in most situations. In this paper, we use physics informed bellman neural network (PIBNN) to solve the optimization problem of the HEV. PIBNN has the ability to obtain the solutions of the Hamilton–Jacobi–Bellman (HJB) partial differential equation with required accuracy. Then, the state and control of the optimization problem of the HEV can be derived by conducting the interference step. Simulation results show the effectiveness and feasibility of the proposed methods.