Improvement Of Adaptive Genetic Algorithm And Its Application In Time Domain Synthesis Optimization Of Pyroshock Response Spectrum

In the design of launch vehicles and satellites,many pyroshock devices are fre-quently used to achieve the separation of loads,the deployment of solar panels and other appendages,and the activation of propellant valves and other systems.These pyroshock devices can generate a pyroshock environment with an extremely high ac-celeration amplitude and a wide frequency range.The shock response spectrum(SRS)is an effective standard tool for engineering analysis and environmental impact quantifi-cation.In existing aerospace structural shock design and test specification documents,only the specification requirements of SRS are given,whereas temporal information is missing.When the equipment or secondary subsystems exhibit a nonlinear dynamical behaviour,the acceleration time history that satisfies the given SRS must be used as the input load for simulation.Therefore,it is very important to synthesize the acceler-ation time domain information satisfying the specification requirements through SRS.To date,some studies have focused on solving this problem.However,most of the existing research methods rely on a considerable amount of experimental data.Synthe-sizing temporal waveforms that satisfy the specified SRS without experimental signals to guide the initial design of spacecraft has become a major problem among engineers.In this dissertation,the related problems of SRS acceleration synthesis are re-searched,including SRS calculation method,SRS characteristic analysis;SRS acceler-ation time domain synthesis method and optimization;improvement of adaptive genetic algorithm.The main contents are as follows:1.Calculation method and characteristic analysis of shock response spectrumOn the basis of literature review,the types and calculation methods of SRS are systematically summarized,and the calculation flow of SRS is defined by typical shock excitation.According to different types of SRS expressions,the characteristics of SRS are analyzed through typical excitation comparison,and some enlightening conclusions are obtained.According to the absolute acceleration SRS most commonly used in the field of aerospace,its characteristics under different shock excitation time,different shock excitation amplitude and different shock excitation waveforms are analyzed.2.A new method for acceleration time domain information synthesis of shock response spectrumIn order to obtain the acceleration time load required for spacecraft structure simu-lation and test analysis,a combined method for pyroshock acceleration synthesis is pre-sented using a series of wavelets at low frequencies and damped sines at medium-high frequencies in the absence of experimental data.The quality of a synthesized SRS is a minimum optimization problem that is evaluated by comparing it with the target SRS.Then the objective function is given,and the basic genetic algorithm(GA)is selected for minimum optimization.Typical examples show that the new method can obtain re-sults closer to the target spectrum than traditional ones,which shows the effectiveness of the new combination method.3.An improved adaptive genetic algorithm with adaptive and nonlinear ad-justments of crossover and mutation probabilitiesAlthough the GA-based optimization results of the new combined acceleration syn-thesis method are better than traditional ones,there is still a certain gap between the final optimization results and the standard target spectrum due to the fixed crossover probability(pc)and mutation probability(pm)values adopted by GA,and there is space for further improvement.Results of the sensitivity analysis of pc and pm values in the optimization of typical example based on GA verify the importance of these two proba-bilities on the final optimization results.In order to further improve the matching degree between the optimization results and the target spectrum,the adaptive genetic algorithm(AGA)with adaptive adjustment of pc and pm is used.A new improved adaptive genetic algorithm(IAGA)is proposed based on the analysis of typical AGAs by standard test functions.The new IAGA is composed of crossover and mutation probabilities with adaptive and nonlinear adjustments,which make the IAGA more stable and easier to find the global optimal solution with fewer evolutionary generations.In the comparison analysis,we demonstrate that combined with the developed IAGA,the new accelera-tion synthesis method can remarkably improve numerical accuracy,and the results are substantially closer to the target SRS than those of traditional methods.4.A novel adaptive genetic algorithm with exponential population size reduc-tion for function optimizationPopulation size is a critical parameter affects performance of AGA.Fixed popula-tion size leads to time complexity,makes the search more complex by increasing gener-ation to converge.Although AGA improves GA' performance,it ignores the influence of population size on optimization results,whose population size is fixed throughout optimization process.To solve the problem,a new Exponential Population Size Reduc-tion(EPSR)strategy is proposed,which uses parameter b to control the population size reduction speed.Furthermore,a noval algorithm termed as AGA-EPSR extends AGA with EPSR is proposed,in which population size reduce at every iteration exponen-tially.Performance of AGA-EPSR is evaluated by optimizing benchmark functions.Results show that AGA-EPSR siginificatily outperforms fixed-size AGA and several well-practiced algorithms like GA,Particle Swarm Optimization(PSO)and Simulated Annealing(SA).