改进响应面法在身管可靠性优化中的应用
Application of the Modified Response Surface Method to Reliability Optimization Design of Gun Barrel

以某速射武器的身管为研究对象,对其烧蚀寿命进行可靠性优化设计,采用改进的响应面模型作为身管烧蚀寿命代理模型。结合遗传算法和高斯牛顿法进行响应面公式参数的拟合。首先用遗传算法求出参数初值,然后高斯牛顿法从这组初值出发迭代搜索响应面模型参数的最优解,最后以得到的响应面模型作为身管烧蚀寿命优化的功能函数进行可靠性优化。该方法充分利用了遗传算法善于进行全局搜索和高斯牛顿法精确的局部寻优能力的优点。结果表明,该方法得出的响应面模型的精度较高,以响应面模型代替有限元模型进行可靠性优化更为方便快捷。
For a certain kind of gun barrel, a modified response surface method to study the reliability optimization design is proposed. To combine Genetic algorithm (GA) and Gauss-Newton method (GNM) is used to acquire the optimal solutions of the model parameters. Firstly, GA is used to solve the initial values of the model parameters. Secondly, GNM is used to get the optimal solution according to the initial values. At last, the reliability optimization design with the agent model can be done. GA is an effective in global solution space searching, while GNM is an effective in local searching. The simulation result shows that the combination of GA and GNM obtains a higher precision of ablation estimation and greatly improves the computational efficiency in the optimization of gun barrel ablation life as well