基于正交试验和GA-SVM的硅胶发泡工艺参数优化

Optimization of silicone rubber foaming process parameters based on orthogonal test and GA-SVM

针对硅胶发泡工艺参数优选时存在的耗时、成本高和准确率低等问题,提出了一种基于遗传算法改进的支持向量机优化方法,该方法在正交试验的基础上利用遗传算法和支持向量机的优点,进行了极差、方差分析,建立了基于遗传算法优化的支持向量机模型GA-SVM,利用该模型对硅胶泡沫材料的表观密度进行了优化,并测试了优化后的硅胶泡沫材料微观结构、力学性能及阻燃性能。结果表明:将正交试验、遗传算法与支持向量机三者结合用于硅胶发泡工艺参数的优化可以明显提高发泡工艺设计效率,GA-SVM优化算法得到的预测值与实测值的相对误差在1.1%以内,且GA-SVM优化算法可获得比单纯使用正交试验更优的硅胶发泡方案,为硅胶发泡工艺参数优化提供了一种新的思路。

Aimed at the problem of time-consuming,high cost and low accuracy of silicone rubber foaming process parameters,an improved support vector machine（ SVM） optimization method based on genetic algorithm was proposed. This method based on orthogonal test utilized the advantages of genetic algorithm and support vector machine,and then analyzed the range and variance of silicone foaming process parameters by using the method of orthogonal test. The GA-SVM optimization model was established to optimize the apparent density of silica foam material with the optimization model. At last,the microstructure,mechanical properties and flame retardant properties of the optimized silicone foam were tested. The results show that the integration of orthogonal test,support vector machines and genetic algorithm has greatly improved the efficiency of foaming process design. The relative error between the predicted value and the measured value obtained by using the GA-SVM algorithm is within 1. 1%.What＇s more,the GA-SVM optimization algorithm can obtain better silicone foaming scheme than only use the orthogonal test,which provides a new idea for the optimization of silica gel foaming process parameters.