Comparison of game theory and genetic algorithm optimisation schedulers for diesel-hydrogen powered system reconfiguration

The turbocharged dual-fuel engine is modeled and connected online to optimiser platform for transient input variation of input parameters decided by designed algorithms. This task is undertaken to enable intelligent control of the propulsion system including the Hydrogen injection instantly to reduce the thermal irreversibility. Therefore, two methods of optimisation are applied to data collected from a turbocharged dual fuel operated propulsion system with direct diesel fuel injection and hydrogen port injection. This study investigates the application of multi-objective game theory (MOGT) and non-dominated sorting genetic algorithm II (NSGA-II) for optimising the performance of a diesel-hydrogen dual-fuel engine. The system is designed in 1D framework with input variability of the turbocharger efficiency, hydrogen mass injection, air compression ratio (Rp), and start of combustion (SoC). The objective is to set maximized the volume work while minimising the entropy generation and NO emission. The first populations in the optimisation procedures are initialised with uniform Latin hypercube and random space filler design of experiment (DoE) for both optimisers. The MOGT can find the best solution faster than NSGA-II with slightly better result. The statistics showed that MOGT generates 12 more unfeasible designs that do not meet the constraint limit on NO emission. The findings indicate that for different optimisation algorithms there are some factors with different effect direction and size on the objectives. Additionally, it is discovered that although MOGT solution makes higher objective function value, the NSGA-II optimal solution leads to better engine efficiency and lower fuel consumption.