Flow control of a proportional directional valve without the flow meter

Flow control is one of the most important valve functions which can be realized using the signal obtained by flow meters. However, flow meters are not suitable for being installed on fast response systems or in compact space due to slow response and large volume. This paper proposes a computational flow feedback control method using a flow inferential measurement method and a proportional integral controller. The flow inferential measurement method uses an AdaBoost neural network to learn flow change regulation affected by spool displacement, pressure differential, and temperature with high accuracy and low overfitting. The flow calculation model is analyzed by global sensitivity analysis to find the effect proportion of each factor, and piecewise wavelet filter is used to reduce the noise effect of the key signal in the calculation. The AdaBoost neural network is found to be capable of recognizing the characteristic shapes of the flow function with an accuracy of ±2%. The controller receives the accurately calculated flow and adjusts the spool displacement to control the flow. The performance of the flow control method is evaluated in the experiments. The control error is controlled within ±3% and the effectiveness of the method is validated.