Multi-objective Optimization Of Automotive Powertrain Based On Adaptive Maintain Diversity Genetic Algorithm

With the rapid development of the world Social economy,the automotive industry has developed rapidly.The automobile has become an indispensable transportation tool in the human life. With the Sharply increase in automobile possession,energy shortages have became is getting worse. Saving energy and reducing consumption in order to improve the automobile transport productivity become is one of the important problem to be solved urgently in automobile industry at present.Whether the matching of automotive powertrain and each part of transmission is reasonable is an important factors affectting automobile power performance and fuel economy.To improve vehicle efficiency and reduce fuel consumption,computer simulation technology is used in this paper,and multi-objective optimization genetic algorithm is used for the best matching of automotive powertrain parameters.To ensure the algorithm to obtain a discrete Pareto solution set of uniform distribution, and has good convergence,this paper puts forward a kind of adaptive diversity keeping strategy,and constructs the adaptive diversity maintaining genetic algorithm for multi-objective optimization (ADMMOGA) based on this strategy.Algorithm which is based on the concept of information entropy gives a "population entropy" used for measuring the population diversity. According to the population entropy of archive population, the algorithm dynamically adjusts the evolution mode of population, realizes the automatic conversion of search mode between "exploitation" and "exploration",effectively controls the number of elite individual retained, explores new individual in sparse areas,guides the composition mode of a new generation population. Because ADMMOGA algorithm has the superperformance in aspects of convergence and distribution, it avoids the stagnation or precocious phenomena generated due to the lack of diversity in others algorithm.The automobile engine and transmission model is set up and the automobile driving equation is given.Engine model includes external characteristics of the engine and model engine fuel characteristics.Automotive power train model includes mathematical model of clutch,transmission model,drive axle and transmission efficiency model of efficiency models. On this basis, optimization of automobile transmission parameter mathematical model of multi-objective optimization is proposed.The file transmission gear ratio and final drive ratio is as for the decision variables.According to the characteristics of different models with different power and economy as a sub-objective function evaluation is selected.The speed ratio and power requirement is as for constraints.In addition,it is also given the automobile power,fuel economy and comprehensive performance of the sub-objective function of the simulation calculation.The dynamic gives the highest speed and acceleration time of maximum degree climbing simulation method.The fuel economy gives fuel consumption per hundred kilometers and more than hundred kilometers driving conditions fuel consumption cycle simulation calculation.The comprehensive performance of the automobile gives drive power loss rate,efficiency of utilization of effective and the simulation of auto calculation of the energy efficiency.A modified difference evolution is proposed in the automotive power train parameter optimization and difference evolution algorithms with double populations are designed for constrained optimization Problems.The paper gives an improved differential evolution based on double populations,improved the defects which the convergence speed is relatively slow in the differential evolution algorithm.The algorithm adopts double populations,one saves feasible solutions,another keep infeasible solutions.It gives two kinds of method for generating new individual,not only keeps good characteristic of the optimal feasible solution, enhances convergence speed,also can effectively use the useful information in infeasible solutions.The algorithm is applied in automobile powertrain system optimization matching, efficiently solve the multi-objective optimization problem of constraints in the processing.The algorithms are used to solve vehicle power train parameter optimization problem for applied research.For automotive power train parameter optimization problem, the encoding algorithm,measure of individual fitness,genetic operators and control parameters of the algorithm is individual specific designed, the constraint is processed on. A mini-car for the calculation is taken as an example.It uses the algorithms for its transmission parameters were optimized.A set of uniform distribution of the global optimal solution is got.The dynamic performance and economic performance the car before and after is juxtaposed.Optimal result is satisfactory.It is verified the proposed algorithm for solving vehicle power train parameter optimization problem is a practical and effective method.