To solve the problems in research and development of bulldozer, a semiphysical experiment system is established to simulate the whole process of bulldozer ope ration. Based on the research results of bulldozer dynamics characteristics, a m odel is suggested and the theoretical foundation of bulldozer automatic control is built. A loading device is used to simulate the force on the bladecontrolle d cylinder. The blade position is adjusted by the blade control system according to the control signal from hardware controller. Then the semiphysical experim ent is performed through simulation calculations. A fuzzy neural network control method based on variable universe is proposed for engine speed control. And the overall control strategy of bulldozer operation efficiency is studied combined with slip rate control and engine management system. Experimental results show t hat the engine speed is stable and the working efficiency is improved when using the new control strategy.