Comparative Study On Multi-objective Optimization Methods Of Methanol Engine

As one of the important research directions of internal combustion engine,the performance optimization of methanol engine has attracted much attention.However,as a nonlinear problem with multi-variable highly coupled and multi-objective mutual constraints,it is difficult to meet the requirements of modern engine optimization by using traditional ways.Therefore,this paper took a methanol engine as the research object,and conducted a comparative study on three multi-variable and multi-objective optimization methods,aiming to promote the development and application of the engine multi-objective optimization method.Firstly,the one-dimensional thermodynamic simulation model of methanol engine was established,and the accuracy of the model was verified through bench test.The influence of intake advance angle,ignition advance angle and air-fuel ratio on engine performance was analyzed to determine the optimization interval of three operation parameters.Then,the mathematical model of methanol engine unconstrained multi-objective optimization was constructed by using the static penalty function method.Based on this,three multi-objective optimization methods were adopted to optimize the three operation parameters,in order to maximize Torque and minimize BSFC.The three methods respectively adopted engine polynomial model,ANN model and GT-Power model to combine NSGA-? to achieve multi-objective optimization.Moreover,two improved methods of initializing population by LHS and crossing dynamically were proposed for NSGA-?,which could effectively improve the optimization effect of the third methods.The results show that the three methods can obtain good optimization results,respectively increasing the Torque by 2.39%,2.22% and 2.24%,and reducing the BSFC by 8.34%,10.11% and 10.05% on average.Finally,the three multi-objective optimization methods were compared and analyzed from the aspects of optimization effect,optimization efficiency and applicability.The results show that: the first method has the advantages of simple application and high optimization efficiency,but its optimization effect is not as good as the other two methods;the third method can obtain good optimization results,but its trial and error cost is high,and the optimization efficiency is about 50% lower than the other two methods;the second method has the advantages of good optimization effect,high optimization efficiency,and wide application,but its modeling process requires higher.In practical application,the methods should be selected according to their characteristics,our own conditions and requirements.