Research on adaptive shift strategy for automatic transmission based on load variation

To address the issues of frequent shifting, significant shift shock, and prolonged power interruption in pure electric vehicles equipped with a two-speed automatic mechanical transmission under dynamic load conditions, this paper proposes an adaptive shift control strategy based on load state recognition. By establishing a Recursive Least Squares-Multi Feedforward (RLS-MFF) estimation algorithm, the characteristics of vehicle load variation are identified in real time. A dynamic load-speed coupling model is constructed to achieve feedforward-feedback composite compensation for target rotational speed. A multi-constraint shift decision mechanism is designed to suppress shift failures caused by abrupt load changes, optimize the control strategy, and improve shift success rates under varying loads. Simulations based on Simulink and bench experiments for two-speed automatic mechanical transmission are conducted. The results show that the load mass estimation error was reduced from ±5.7% with traditional methods to ±1.16%, significantly improving dynamic load identification accuracy. Gear shift frequency was reduced by up to 17.6%, effectively suppressing the phenomenon of frequent shifting and resolving the issue of frequent gear changes under dynamic load conditions. This provides theoretical foundation and engineering insights for enhancing the control performance of multi-speed transmissions in pure electric vehicles.