Abstract: A dead-section and integral saturation restrain PID controller is designed according to the character of the hydro-mechanical transmission. BP neural net is used to adjust the three parame-ters of the PID control arithmetic using the most-rapid descending arithmetic with a inertia item. The number of medial node in the BP neural net, the study rate and inertia coefficient in the most-rapid descending arithmetic are de-termined by theoretical analysis and tests respectively. The results of the tests in the lab show that the PID controller can make the output of the driving unit be constant under speed adjusted. That means the PID controller can make the engine work in the most powerful or the most economical mode by adjusting the ratio of the hydro-mechanical transmission if that is used in the automobiles.

Key words: Hydro-mechanical transmission, Neural net, PID control, Stepless transmission