Predicting the performance of an electrohydraulic servomotor using mathematical modeling and neural networks

This paper investigates the potential of using the Mathematical Modeling and Neural Networks techniques in modeling nonlinear systems whose dynamics are important. The electrohydraulic servomotor is the system under study in this paper. The two techniques were applied to model the servomotor system. The developed Mathematical Model takes into consideration the basic servomotor nonlinearites. The developed Neural Networks model uses various optimization methods in the training process. Measurements of the steady state and dynamic performance of the system were carried out and they were used to evaluate the deduced models. A computer simulation, based on the developed models, was performed to predict the steady state response and dynamic performance. The obtained results were compared with the measured ones. In the steady state mode of operation, the difference between the experimental and the predicted results is almost zero and the two techniques predict almost the same results. In the transient mode of operation, the Neural Networks may be used to model highly nonlinear systems accurately; however, it requires a large quantity of experimental data describing the system behavior in the training process