基于DOE设计与MOPSO算法的PVC建筑排水管多目标优化分析

Multi-objective Optimization Analysis of PVC Building Drainage Pipe Based on DOE Design and MOPSO Algorithm

以建筑排水管为研究对象,采用DOE设计中的最优拉丁超立方抽样法随机均匀的抽取样本,以吹胀压力、吹胀时间、型坯初始温度以及型坯初始壁厚为设计变量,以壁厚均匀性函数值为响应目标,建立代理模型以替代CAE的模拟仿真。利用多目标粒子群优化算法(MOPSO)对代理模型进行内部寻优,从而获得较为理想的成型工艺参数。结果表明:吹胀压力0.6 MPa、吹胀时间3 s、型坯初始温度169.09℃以及型坯初始壁厚2 mm时制件成型质量较为理想。该工艺条件下,壁厚均匀性函数值实际模拟值为0.32018×10^(-5)mm^(2),预测值为0.3199×10^(-5)mm^(2),两者相对偏差为0.087%,小于5%。

Taking the building drainage pipe as the research object,the optimal Latin hypercube sampling method in the DOE design is used to randomly and uniformly sample samples.The inflation pressure,the inflation time,the initial temperature of the parison and the initial wall thickness of the parison are used as design variables.The wall thickness uniformity function value is the response target,and a proxy model is established to replace the CAE simulation.The multi-objective particle swarm optimization algorithm(MOPSO)is used to optimize the agent model internally to obtain ideal molding process parameters.The results show that the molding quality of the parts is ideal when the inflation pressure is 0.6 MPa,the inflation time is 3 s,the initial temperature of the parison is 169.09℃,and the initial wall thickness of the parison is 2 mm.Under this process condition,the actual simulation value of the wall thickness uniformity function value is 0.32018×10^(-5)mm^(2),and the predicted value is 0.3199×10^(-5)mm^(2).The relative deviation between the two is 0.087%,which is less than 5%.