Aiming at the disturbance suppression of nonlinear friction and the output feedback control problem, this paper presents a high order sliding mode extended state observer(ESO) to obtain real-time state signals. On this basis, an adaptive neural update law is designed, obtaining the relationship between the control signal parameters setting and output tracking performance. Meanwhile, the necessary and sufficient conditions of observer state initial value and adaptive parameter initial value are given to guarantee system dynamic performance. Finally, the stability proof and simulation results verify the effectiveness of the proposed control algorithm.