Modelling and control of proportional directional control valve using neural network

Abstract EN

A multi-layer, multi-input/ multi-output feed-forward back propagation neural network will be used to identify the model of proportional directional control valve with the addition of its non linearities.

An electro-hydraulic training test bench is used to collect data for training the neural network.

Using MATLAB, SIMU1LNK the electro-hydraulic controlled system is tested, by applying a variable reference stroke position.

A PI controller is tuned to get a stroke response with minimum overshoot and minimum steady state error.

Modeling results were very satisfied and were very close to the experimental one with an error less than 10'10.

This work can be generalized by applying the same idea to any nonlinear valve.