摘要弹丸起始扰动与火炮射击密集度密切相关,具有随机性、不确定性的特点。本文以提高火炮射击密集度为研究目标,运用优化设计与稳健设计相关理论,对面向密集度的弹丸起始扰动进行探索研究。应用蒙特卡洛模拟构建射击密集度模型,并采用FORTRAN语言编写密集度计算程序。在多学科优化软件ISIGHT环境中集成密集度计算程序并构建神经网络模型。进行灵敏度分析,量化弹丸起始扰动各参数对射击密集度的影响程度。从中选出对密集度影响较大的参数作为优化设计变量,采用多目标优化方法改进非支配排序遗传算法(NSGA-II),进行弹丸起始扰动参数优化,得到全局最优解与合理容差使纵向射击密集度达到预期目标。采用稳健设计方法在弹丸起始扰动参数最优方案基础上进行稳健设计,使得最终的起始扰动参数方案既满足目标要求又具有较好的稳健性。33345
关键词:射击密集度,弹丸起始扰动,神经网络,灵敏度分析,遗传算法,稳健设计 毕业论文设计说明书外文摘要
Title Optimization design and robust design of the projectile initial disturbance for firing intensity
Abstract
The projectile initial disturbance is closely related to the intensity of the howitzer, random and uncertain. This thesis focus on improving the intensity of the howitzer based on the theory of optimal design and robust design. The FORTRAN program language is used to construct shooting intensity model, which simulates randomness of the projectile initial disturbance and based on the related theory of Monte Carlo simulation. The program is integrated into multidisciplinary optimization software (ISIGHT), and a neural network model of firing intensity is built. A quantitative description about the influence of various parameters on the fire intensity is given through sensitivity analysis. The projectile initial disturbance parameters are optimized,the global optimal solution and reasonable tolerance is got by adopting the multi-objective optimization method (NSGA II), which can make the enfilade intensity to reach the expected aim. A robust design is performed to make the final parameter scheme satisfy the requirement of target and have good robustness based on the optimal solution.
Keywords: Firing intensity, projectile initial disturbance, neural network, sensitivity analysis, genetic algorithm, robust design.
目 录
1 绪论 1
1.1 选题背景与意义 1
1.2 国内外研究现状 1
1.3 本文研究内容 4
2 火炮射击密集度模型 5
2.1 射击密集度流程 5
2.2 随机变量及随机分布的确定 5
2.3 随机抽样方法 7
2.4 射击密集度计算程序 8
2.5 本章小结 10
3 ISIGHT环境下射击密集度神经网络近似模型构建 11
3.1 ISIGHT与射击密集度程序的集成 11
3.2 人工神经网络理论 14
3.3 ISIGHT中神经网络近似模型 16
3.4 本章小结 20
4 弹丸起始扰动参数灵敏度分析 21
4.1 试验设计方法 21
4.2 初步试验设计与灵敏度分析 22
4.3 基于神经网络近似模型的灵敏度分析 27
4.4 本章小结 30
5 弹丸起始扰动参数优化设计 31
5.1 改进非支配排序遗传算法 31