Abstract:

In order to fully explore the energy saving potential of plug-in hybrid electric vehicle with determined power parameters of powertrain and control strategy, and in meeting the vehicle performance constraints, the running cost of 100 km was proposed as the response. The optimal Latin hypercube design was used to investigate the spatial response characteristics of all factors of the transmission system, based on which the key design factors of the system were identified and the optimal local area was confirmed. Combined with making full use of the advantages of strong local optimization ability of nonlinear quadratic programming algorithm, the Radial Basis (RBF) neural network model was established to obtain global optimal solution and the best response of the optimization problem. The model error is 1.06% which shows that the optimization method based on approximation model is of high accuracy. The running cost of 100 km is reduced by 9.72%.

Key words: vehicle engineering, plug-in hybrid electric vehicle(PHEV), running costs of 100 km, optimal latin hypercube design, RBF neural network, nonlinear quadratic programming algorithm