Research On Optimization Methods Of Flexible Multi-body Dynamics Oriented To Integration Design Of Large Calibre Artillery Gun

With the continuous change of war form,higher requirements on the performance of the artillery gun are put forward for the new battlefield environment and combat mode,so the contradiction between power and mobility of large caliber artillery gun is increasingly outstanding.However,by traditional artillery gun design theory,it is difficult to achieve the optimization of kinetic parameters which affect the general properties,thus it is unable to crack this contradiction,it has become one of the main bottlenecks of developing a new generation high-performance artillery gun.Based on this background,with a large caliber towed artillery guns as the research object,structural optimization technology of flexible multi-body system dynamics is studied with integrated use of multi-body system dynamics theory,neural network surrogate model technology,and modern optimization design theory and method.In order to better simulate the mutual coupling of the space motions and local deformation of large caliber artillery's different parts in the firing process,modal synthesis method is used to describe the elastic deformation of key components,which are tube,cradle,and upper carriage.And the above flexible bodies are connected with rigid bodies,like the breeches and the muzzle brake,through interface nodes.A modified Hertz contact model and a modified Coulomb friction equation were introduced,and using Flex-Flex contact model to simulate the relationships between bushings and barrel.Launch dynamic test data were utilized to verify the rationality of the dynamic model.In allusion to the problem that the flexible body structural parameters of implied in the modal neutral file and difficult to be chosen as design variables directly,a GA optimized RBF-BP neural network model is proposed to simulate the nonlinear relationship between the artillery flexible structural parameters and the optimization goals,then combined with NSGA-? genetic algorithm,the flexible body structure achieved optimization.With a large caliber artillery gun as an example,the GA-RBF-BP neural network optimization model of artillery launch dynamic is established,the muzzle vibration characteristics as optimization objectives,and nondominated sorting genetic algorithm is used to find the Pareto front.The max-min criterion is adopted to select a compromise structure scheme from the Pareto front.To solve the gun structure performance optimization problems about large computational cost,slow convergence and fall into local optimum easily,the local and global resampling strategy are introduced,an artillery gun flexible multi-body structure optimization method based on the adaptive RBF neural network is proposed,and three optimization examples are tested to verify the feasibility of the proposed optimization method.New samples are added according to the local and global resampling strategies,thus guiding the optimization process fast convergence to global(local)optimum solution in the design of artillery gun.For the multi-objective optimization problem,physical programming method is utilized to transform the multi-objective problem into a single objective optimization problem,penalty function method to solve the constraint problem,and genetic algorithm to obtain the current optimal solution.By solving the Sphere functions and Ackley functions,the optimization method based on the adaptive RBF neural network is tested.Compared with gun barrel structural optimization only using the genetic algorithm,and with the muzzle disturbance optimization in chapter 3,the optimization results in the second and third optimization examples show that the optimization method has higher optimizing efficiency and better optimization effects.In view of that,in the present artillery structure optimization,the global parameters optimization and the key components optimization are dealt with separately,an integrated optimization method has been proposed.Considering the structure stiffness constraints,the muzzle disturbance and firing stability are set as the optimizationobjecties,general parameters and key structural parameters of tube and cradle selected as design variables,adaptive RBF neural network utilized as optimization method,an integrated optimization design of the big caliber artillery gun is carried out.Results show that,at the instantaneous moment when the projectile comes out of the tube,the transverse angular displacement,vertical angular displacement,vertical angular velocity in the center of the muzzle has been decreased significantly.Also,for the entire period of the firing process,the maximum horizontal displacement of the trail spade centre and the maximum vertical displacement of the front seat plate centre reduced observably.These mean that the muzzle disturbance effectively decreases,and the stability of the howitzer improves a lot.