Numerical Comparison of PI and Neural Network-Based Controllers for the Hydrostatic Unit in Hydro-mechanical Transmissions of Self-propelled Vehicles

The application of hydro-mechanical transmissions is recently the trend in agricultural vehicles where a continuously variable transmission ratio has advantages. Hydro-mechanical transmissions provide efficient power transfer while maneuverability is still maintained, and therefore, fuel efficiency is enhanced. Nevertheless, the main issue in their employment is a precise control of the hydrostatic unit, whose physical characteristics are highly nonlinear and affected by unknown disturbances. In order to exploit the advantages of the system, the transmission ratio of the hydrostatic unit needs to be controlled properly to maintain the optimal working point of the internal combustion engine (ICE). This article presents numerical comparison results of a proportional-integral (PI) and a neural network (NN) based controller applied to the hydrostatic unit of a hydro-mechanical transmission system, which was designed to be deployed on a self-propelled agricultural vehicle. The controls were established in a discrete-time domain aiming at a practical outcome, where the control algorithm could be implemented on an industrial computer to perform the control tasks.