The powertrain efficiency of the engineering vehicle under the heavy load is decreased obviously. In order to solve this problem, the method of parallel self-adaptive neural network was employed based on “two parameters” shift schedule, which of structure include neural network control, self-adaptive neural network model, network evaluation and running monitor model. The simulation results showed that the intelligent shift control could improve the powertrain efficiency of the engineering vehicle, and could overcome the low realtime behavior of neural network in the meantime.