论文部分内容阅读
将混合优化算法引入粒子分离器优化设计,减小粒子分离器流动损失并简化结构.利用四次样条曲线参数化描述粒子分离器模型,建立粒子分离器自动化仿真流程.采用优化拉丁方实验设计方法获取样本点并建立椭球径向基函数神经网络代理模型,基于该模型使用非支配排序遗传算法进行全局多目标优化,最后采用序列二次规划算法进行局部优化,得到了粗砂分离效率达到100%、细砂分离效率达到86.7%,总压损失小于0.6%、出口总压畸变较小的设计方案.
The hybrid optimization algorithm was introduced into the optimization design of particle separator to reduce the flow loss and simplify the structure of the particle separator.The quadratic spline curve parameterization was used to describe the particle separator model and to establish the particle separator automation simulation.Using the optimized Latin square experimental design Methods The sample points were obtained and the radial basis function neural network proxy model of ellipsoid was established. Based on the model, a non-dominated ranking genetic algorithm was used to optimize the global multi-objective. Finally, the quadratic programming algorithm was used to optimize the local optimization. 100%, sand separation efficiency of 86.7%, the total pressure loss of less than 0.6%, the total export pressure distortion design.