Electro-Hydraulic Proportional Position Control Using Auto Disturbance Rejection Based on RBF Neural Network

Large friction force and large dead zone are two typical nonlinear characteristics of electro-hydraulic proportional valve controlled hydraulic cylinder position control system. Aiming at those characteristics, a dead zone dynamic compensation algorithm is researched in order to reduce the lag time and control error. At the same time, a control strategy of radial basis function (RBF) neural network combined with auto disturbance rejection control (ADRC) is researched according to the impact of different conditions. The experimental result shows that the proposed algorithm improves performance of the electro-hydraulic proportional valve controlled hydraulic cylinder position control system. In positioning control experiment, the overshoot is 0 and the stability error is 0. In tracking control experiment, the lag time is reduced from the original 1.5 s to 0.2 s with no flat top phenomenon and the maximum error was reduced from 20 mm to 3 mm.