كامبيز نيكان

Optimal gear shifting strategies in automatic transmissions affect fuel consumption and drivability. In general, automatic transmission gear change is controlled based on two-dimensional lookup tables, which are measured and used based on experimental results in a steady state. Tolerances and component wear can cause factory measured shifting to have deviations that negatively affect clutch durability and driver comfort. This research proposes an optimal gear shifting strategy with the help of short and long term memory algorithm. Unlike the conventional methods that predict the vehicle speed factor to optimize the gear shift with the help of modeling the power transmission system, in this method, using a multi-variable short-term long-term memory algorithm, consisting of vehicle speed, throttle, and vehicle acceleration parameters. And the data of the car's route finder intelligently predicts the appropriate gear change time. The proposed strategy is through simulation of driving cycles in AVL Cruise software and artificial neural network methods and deep learning in other articles and in the range are eva‎luated at different times. The obtained results show that in addition to increasing the accuracy of gear change prediction from 94.4% to 99%, we see a reduction in the number of gear changes and the elimination of sudden gear changes using the proposed method, which can cause Increase driving ability and fuel consumption efficiency compared to the base model in cruise software up to 7.7%.

گيربكس اتوماتيك , استراتژي تعويض دنده , تعويض هوشمند دنده , يادگيري عميق

automatic gear box , gear shifting strategy , smart gear shifting , deep learning

kambiz nikan

DR. behrooz mashhadi