Parameter Optimization Of Hybrid Vehicle Control Strategy Based On Game Evolution Algorithm

Vehicles powered by fuel have brought mankind conveniences and economic interests, but it also lead to the serious energy shortage and environment pollution problems. Development and application of new energy vehicle has become a hot topic and megatrends around the world.Hybrid Electric Vehicle(HEV) has become a developmental focuses in national governments and automobile industry at the present stage because of its advantages in the energy conservation and environmental protection as well as the outstanding power performance. Control strategy is a very important part of HEV, which directly influences the performance of the vehicles. Excellent vehicle control strategy is capable of improving the vehicle power system's work efficiency and reducing fuel consumption and pollution emissions and satisfying the requirements of dynamic performance by scheduling the operating state of difference power sources like engine and battery according to real-time driving cycle, and balancing the energy flow among the subsystems. Focus on the optimization problems of HEV control strategy, game evolutionary algorithm based on behavioral game theory(Game EA) was designed as a new effective optimization method for solving the problem. The main research contents in this paper are as follows:(1) The paper summarized the current optimization method for the HEV control strategy,stated briefly the typical structure and control strategy of HEV, depicted the simulation method of HEV, and introduced ADVISOR simulation software of HEV. The vehicle simulation model of parallel HEV and some main sub-modules in the ADVISOR were also illustrated and analyzed.(2) This paper introduced the basic knowledge and theory of game theory, studied the behavior game mechanism of game theory. A game evolutionary algorithm based on behavioral game theory(Game EA) was put forward, and the individual learning frame of imitation and belief learning are designed. The individuals make decisions by checking the payoffs expectation, and the imitation operator is used to revise gene so as to learn from other competitor, and the belief learning operator is employed to mutate chromosome to improve competitiveness. The results on thirteen benchmark problems show that Game EA outperforms the comparison algorithms on accuracy and exploration.(3) Game EA was applied to optimize the HEV control strategy. Considering the acceleration,maximum gradability, and top speed of HEV as the constraint conditions, the minimization of the sum of fuel consumption(L / 100 km), HC(g/km) and NOx emissions(g/km), and CO emissions(g/km) was used as the optimization goal. The Matlab and ADVISOR were employed to test the performance of HEV. The simulation experimental results of 14 runs show that the fuel consumption of per hundreds kilometers, HC and NOx emissions, CO emissions was down by at least 9.46%, 35.88%, 37.93% than the old system. The engine, motor and overall system efficiency were at least increased by 12.5%, 44.44%, 18.39% respectively.