Research On Particle Swarm Optimization With Alpha-stable Mutation And Its Application In DSI Aerodynamic/Stealth Optimization

The rapid development of aviation technology has put forward higher requirements for the performance of modern aircraft.It makes the way of designing modern aircraft has a great difference from the traditional.Such as efficient aerodynamic characteristics,good stealth performance,high reliability of the control system and advanced integrated avionics system and so on.Especially in aerodynamic/stealth design.Aerodynamic design determines the basic characteristics of the aircraft and directly affects the other systems.And stealth design requirements may overturn the original aerodynamic shape.Therefore,the design of modern aircraft needs to take both into consideration and balance each other.The best solution to realize this process is to establish an optimal aerodynamic stealth design system,which is optimized by computer.However,the current aerodynamic stealth optimization design still has many shortcomings.One of the most critical problems is reflected in the optimization algorithm,that is,the optimization efficiency is low,the accuracy is not high and the scope of application is small.The traditional particle swarm optimization algorithm is analyzed in detail.By introducing the new mutation mechanism,the single-objective particle swarm optimization algorithm,the multi-objective particle swarm optimization algorithm and the multi-objective particle swarm optimization algorithm based on the proxy model are improved and established.The application effect of a certain inlet model in aerodynamic and stealth optimization design was verified.The main work of this paper is as follows:1.A new variation method based on Alpha stable distribution is proposed.A dynamic variation strategy to adjust the stable coefficient of Alpha stable distribution is proposed in this paper.In the initial stage of Particle Swarm Optimization(PSO),a small steady-state coefficient is used to generate variation,which can enhance the diversity of population and reduce the possibility of falling into local optimization.At the end of the optimization algorithm,a larger steady-state coefficient was used to generate a variation method similar to the Gaussian distribution,which enhanced the local search ability of PSO algorithm and improved the accuracy of the solution.The Alpha Stable Particle Swarm Optimization(ASPSO)algorithm based on Alpha stable variation is compared with the basic PSO and Differential Evolution Algorithm(DE).The test results show that the new ASPSO algorithm greatly improves the convergence speed and accuracy of the algorithm.In the single point transonic drag reduction optimization of RAE2822,the optimization effect and efficiency of ASPSO algorithm are much higher than that of traditional PSO algorithm when the population size and other parameters are kept the same.ASPSO optimized airfoil also has higher lift-drag ratio than PSO optimized airfoil.2.The multi-target particle swarm optimization(PSO)algorithm is analyzed and improved according to the multi-target situation in aircraft aerodynamic stealth optimization.Based on the Crowding Distance Multi-objective Particle Swarm Optimization(CDMOPSO)algorithm and combining with Alpha stable mutation,a new multi-objective particle swarm optimization algorithm ASMOPSO was constructed.The algorithm uses an external file to store the non-dominant solutions found by PSO.For the limitation of external archive capacity,ASMOPSO used the crowding distance operator to reduce and maintain the external archive.Meanwhile,the crowding distance was also used to select the globally optimal particle from the external archive to guide the population to move to the real frontier.The Alpha stable mutation operation improved the population diversity of MOPSO and greatly improved its ability to solve multi-modal and multi-objective problems.The classical test functions such as ZDT series are used to compare and verify the optimization effect of the algorithm before and after the improvement.The results show that the improved ASMOPSO algorithm is highly competitive in solving the optimization problem of ZDT4,a super multi-local frontier.RAE2822 airfoil multi-objective optimization results show that the multi-objective frontier generated by ASMOPSO is closer to the real Pareto frontier,which is obviously better than the CDMOPSO algorithm.The simulation of aerodynamics and stealth of three-dimensional aircraft depends heavily on computer resources.It needs to call real functions many times and consume a lot of computing time.Furthermore,the efficient multi-objective particle swarm optimization algorithm based on the expect hyper-volume improvement(EHVI)infill criterion is studied.On this basis,a new EHVI-ASMOPSO optimization algorithm suitable for real engineering optimization needs is established.In the multi-objective optimization process,the Kriging surrogate model which based on the expect hyper-volume improvement(EHVI)infill criterion is adopted for integrated optimization design.The EHVI value of the population was calculated by combining with the Pareto frontier obtained by ASMOPSO.Select several individuals with the largest EHVI value to evaluate the real function and add sample points while updating the external file.This method reduces the number of calls to the real objective function during each PSO loop and greatly improves the optimization efficiency.In the test function comparison,the EHVI-ASMOPSO optimization algorithm has better optimization performance than the EHVI-CDMOPSO optimization algorithm under the same parameter setting condition.Especially in ZDT4,it has excellent convergence ability and shows very high optimization efficiency.3.Aiming at the requirement of electromagnetic stealth calculation,the numerical calculation method of electromagnetic stealth based on MLFMM,PO and LEPO was studied.Compared with the experimental data,the advantages and disadvantages of the three electromagnetic stealth values are analyzed.Finally,the LEPO method combined with UTD theory was chosen as the electromagnetic stealth characteristic calculation method.The efficient multi-objective particle swarm optimization algorithm based on EHVI is applied to the aerodynamic/stealth multi-objective optimization design of the flying wings UAV.The FFD parametric method was used to parameterize the wing layout.According to the comparison of airfoil profile calculation and RCS calculation results,the optimized UAV has better aerodynamic performance and stealth performance.4.The bump surface of DSI was taken as the research object to carry out the research on the efficient multi-target aerodynamic and stealth optimization design of DSI inlet.The improved EHVI-ASMOPSO algorithm was selected to verify and analyze the optimal design of aerodynamic and stealth characteristics of the DSI inlet.Firstly,by comparing the calculation results of aerodynamic characteristics of different turbulence models in the inlet,the k-ω SST model was selected to optimize the design.The calculation method of electromagnetic stealth combined with LEPO and UTD was used to evaluate the stealth characteristics of RSC.The FFD parametric method is used to parameterize the bump surface.Subsonic and supersonic flight conditions were selected for the comprehensive optimal design of DSI inlet.The total pressure distortion coefficient of the inlet and the mean value of forward RCS in the supersonic design condition were taken as the optimization objectives.The total pressure recovery and mass flow rate in subsonic and supersonic design conditions were taken as constraints.The optimization results show that the improved algorithm achieves a better optimization scheme with fewer calls to the objective function.Although the distribution of the optimized Pareto solution is sparse,the distribution range is wide.Compared with aerodynamic characteristics,the optimization degree of stealth is smaller.This also indicates that bump surface has a more significant impact on aerodynamic characteristics of the inlet at the same deformation amount.