Adaptive Feedforward Compensation Using a Neural Network for Velocity Control of Hydrostatic Transmissions

This paper presents study results of a model-free approach towards velocity tracking control for a hydrostatic transmission system. In a decentralized control structure, the control of the hydraulic motor bent-axis angle is designed in a feedforward manner based on the desired trajectory of the system output, whereas the tracking control of hydraulic pump swash-plate angle for tracking the desired output value relies on an adaptive feedforward error compensation scheme using a multi-layer perceptron neural network. The proposed control structure has been evaluated in simulations and validated by experiments on a dedicated test rig at the Chair of Mechatronics, University of Rostock.