Electro-hydraulic piston control using neural MRAC based on a modified state observer

A new model reference adaptive control design method using neural networks that improves both transient and steady stage performance is proposed in this paper. Stable tracking of a desired trajectory can be achieved for nonlinear systems having significant uncertainties. A modified state observer structure is designed to enable desired transient performance during uncertainty learning. The neural network adaptation rule is derived using Lyapunov theory, which guarantees stability of the error dynamics and boundedness of the neural network weights. An extra term is added in the controller expression to introduce a 'soft switching' sliding mode that can be used to adjust tracking errors. The method is applied to control the velocity of an electro-hydraulic piston, and experimental results demonstrate the desired performance is achieved with smooth control effort.