Modeling the efficiency characteristics of hydraulic pump via neural networks

After the efficiency characteristics of hydraulic pump are analyzed, the modeling method for the efficiency characteristics of hydraulic pump is studied based on neural networks. The piston pump with displacement 39.57 cm³/r is used as the modeling object, and a few of experimental data are used as the training examples of neural networks. The networks are trained by using three-layer BP neural networks and Bayesian regularization, and the flow and torque of the hydraulic pump are modeled. The total efficiency of the hydraulic pump is calculated on the basis of model of flow and torque. The experimental data obtained from different velocity and pressure working conditions are used to verify the extensive ability of the model. The studied results show that the total efficiency errors of the hydraulic pump are less than 0.2%.