Multi-Objective Genetic Algorithm for an automatic transmission gear shift map

Reductions in engine emissions including carbon dioxide (CO2) can be achieved by analysing the Brake Specific Fuel consumption (BSFC) map and selecting a set of gear ratio and associated gear shift map which will move the engine operating point into the most efficient region. This paper formulates the determination of the most appropriate gear shift map for a Dual Clutch Transmission as a multi objective optimisation problem. A multi-objective genetic algorithm (MOGA) is complemented by a new repair mechanism to enforce acceptable driveability in the solutions produced as well as a new problem specific local search operator exploiting engineering knowledge to determine feasible gear shift maps that result in reduced emissions. The solutions are evaluated against standard as well as novel objective formulations that take into account CO2, driveability, gearbox durability and other practical implementation considerations. A correlation analysis, combined with Pareto analysis, is performed to identify and visualise related objectives that can be combined using a traditional weighted sum approach and those that should be handled separately using a Pareto front approach to facilitate the trade-offs identification and resolutions. The first set of objectives aim to reduce CO2, the second to maintain driveabil-ity, and the third to consider durability. The novel approach was used to identify trade-off solutions focusing on reducing CO2 emission. These solutions were validated on a prototype vehicle on a rolling road using the New European Driving Cycle, confirming the expected CO2 savings whilst conserving good driveability characteristics.