应用遗传算法优化支持向量回归机的喷气涡流纺纱线质量预测

为探究熟条质量对喷气涡流纺纱线质量的影响，建立了遗传算法优化的支持向量回归机预测模型。模型的输入端参数为熟条的4项指标（条干CV值、回潮率、定量和定量不匀率），分别对19.7tex 和11.8tex的涤纶/粘胶（67/33）喷气涡流纺纱线进行强力和条干CV值预测试验，同时建立了BP神经网络模型作对比试验。2种模型预测对比分析的结果表明：遗传算法优化的支持向量回归机模型的稳定性和精度要比BP神经网络模型高得多，更适用于描述熟条质量与喷气涡流纺纱线质量（单纱强力和纱线条干CV值）间的非线性关系。

Prediction of vortex yarn properties based on hybrid genetic algorithm and support vector regression

Abstruct: In order to make a primary research in the relationship between the quality of drawing sliver and the quality of vortex spinning yarn. Support vector regression machine prediction model optimized by GA is built up. 19.7tex and 11.8tex vortex spinning blended yarn of polyester and viscose（the blending ratio is 67:33）is selected as the experiment object. Yarn strength and CV of yarn unevenness are predicted while four quality parameters of drawing sliver (CV of yarn unevenness, moisture regain, quantification of sliver and unevenness of quantification) are used as the input parameters of prediction model. BP neural network model is also built to make a comparison with the aforementioned model. After a comparison between these two models, the result shows that the model of the optimized support vector regression machine performed a more powerful reliability and accuracy and it can describe the non-linear relationship between the quality of sliver and the quality of vortex spinning yarn more appropriately than BP model.