Endurance Brake Classification Control Based on the Generalized Growth and Pruning Radial Basis Function Neural Network

Abstract

Because of the difference of working principles and arrangements of endurance braking systems, including engine brake, exhaust brake and eddy current retarder, it is difficult to match braking manually more than two types of endurance braking systems working simultaneously on long downhill. Meanwhile, manipulating control on different slopes will distract the driver's attention and cause driving fatigue. Aiming at this problem, the endurance brake classification control strategy is proposed, setting the deceleration, road slope and the difference of current speed and target speed as an input and the endurance brake classification as an output variable. Considering velocity variation is related to these factors with strong nonlinear characteristics, Generalized Growth and Pruning Radial Basis Function neural network control is used to estimate the input deceleration. Tests were conducted to verify the accuracy of simulation model. Variable slopes are researched through simulation method. The results show that the system designed to achieve automatic matching control can effectively decelerate and keep the truck running stably.