Gear Predictor Of Manual Transmission Vehicles Based On Artificial Neural Network
Free (open access)
37 - 46
A. Gardel & D. Pérez
Nearly all mechanical systems involve rotating machinery (i.e., a motor or a generator), with gearboxes used to transmit power or/and change speed. Concerning vehicles, there is a specific nonlinear relationship between the size of the tires, linear velocity, engine RPM, gear ratio of the differential, and the gear ratio of the transmission. However, for each car there is a specific range of gear ratio of the transmission for each gear. On the other hand, the gear value is an indication of the driver behaviour and the road conditions, therefore it should be considered to establish non-pollutant driving guidelines. In this paper, two novel feed-forward artificial neural network (ANN) models have been developed and tested with the gear as the network output and the velocity of the engine (RPM) and the velocity of the car in (Km/h) as the network inputs. A lot of experiments were made using two commercial cars. The prediction efficiency of the proposed models is superior (i.e., the generalization mean square error is about 0.005). However after testing with two different vehicles, the conclusion is that on one hand the structure of the ANN model is suitable. On the other hand each vehicle has its specific model parameters. This paper shows that it is difficult to develop a universal model that predicts the gear based on the RPM and speed of any car. Keywords: feed-forward artificial neural networks, gear predictor, manual transmission.
feed-forward artificial neural networks, gear predictor, manualtransmission.