Modeling and predictive analysis of the hydraulic GEROLER motor based on artificial neural network

Abstract

GEROLER hydraulic motors are known for their good value for money and their balance between simplicity, robustness, compactness, versatility and noise. Compared to axial hydraulic motors, GEROLER motors still represent a research area with the possibility of a significant contribution in terms of nonlinear dynamic behavior analysis. The aim of this research was experimental analysis of GEROLER motor dynamics at uneven load torque. Based on the obtained laboratory measurements, a black-box model for predicting the operating parameters using the artificial neural networks was developed. Two different neural network architectures were used: the simpler static multilayer feed-forward network and the more complex dynamic NARX neural network. From the obtained results, it appears that the multilayer feed-forward neural network provides acceptable results, while the dynamic NARX neural network provides more favorable results due to its flexibility in dealing with nonlinear dynamic systems. The research conducted represents a new approach for modeling and predictive analysis of the GEROLER engine.