Abstract: Proportional servo valves are widely applied in intelligent engineering machinery, national defence equipment and other high-end hydraulic systems. For the intelligent proportional servo valve, flow force is the most important factor limiting the improvement of its intelligent level. In order to solve the above problems, a flow force compensation neural network-based robust controller(FF-NNRC) of the valve centre is developed. Firstly, Fluent is employed to obtain the flow force information of proportional servo valve under different spool displacements and pressure boundary conditions, which can be used to simulate the flow force disturbance of practical working conditions. Neural network is designed to learn and approximate the flow force disturbance, then handles it in the feedforward model compensation term dynamically, robust term is formulated to deal with other disturbances and neural network estimation error. Lyapunov stability theory proves that the proposed control strategy can achieve the bounded stability of the system. Simulation results show that, compared with traditional PID controller and model compensation robust controller(MC-RC), the proposed controller has higher control accuracy and anti-interference ability.

Key words: flow force, computational fluid dynamics(CFD), neural network, model compensation, robust control