摘要
根据经实验验证的玻璃钢(GFRP)拉挤工艺过程数学模型,以数值模拟结果为样本数据,建立反向传播(BP)神经网络,得到拉挤工艺参数(固化温度、拉挤速度)与GFRP固化度间非线性相关关系。采用神经网络结合带精英策略的快速非支配排序遗传算法(NSGA-Ⅱ)解决拉挤过程中固化炉温度和拉挤速度多目标优化问题,得到了拉挤优化问题的Pareto最优解集。实验结果表明,优化后的工艺参数能有效提高生产率,降低固化炉温度,效果显著。
The back propagation（BP） neural network was trained to form the relationship between the technological parameters（die temperature,pull speed） and the degree of cure on the basis of the simulated results,which were predicted by the mathematic model for glass fiber reinforced plastics（GFRP） pultrusion.The fast elitist non-dominated sorting genetic algorithm（NSGA-Ⅱ） was adopted to deal with the multi-objective optimization for GFRP pultrusion,and the Pareto solutions for the optimization problem were found out.It shows that there was a significant improvement for technological parameters after optimization.
出处
《玻璃钢／复合材料》
CAS
CSCD
2010年第5期57-61,共5页
Fiber Reinforced Plastics/Composites
基金
黑龙江省自然科学基金项目(E01-10)
关键词
玻璃钢
拉挤
数值模拟
神经网络
NSGA-Ⅱ
多目标优化
glass fiber reinforced plastics
pultrusion
simulation
neural network
NSGA-Ⅱ
multi-objective optimization
