Considering the symmetric valve controlled asymmetrical cylinder electro-hydraulic servo system of nonlinear and in order to improve the control precision of the system, a novel feedback linearization control strategy is put forward based on the analysis of the inherent property of the system.The feedback linearization stability of the system is analyzed, and adaptive compensation control for uncertain parameters is introduced by using the neural networks. Finally reasonable parameters are chosen for the simulation.Experimental results show that the neural network technique based feedback linearization valve control system not only has good stability, high control accuracy, also the system tracking performance is superior, which is applicable to real time control situation.