用临近支持向量机预测熔喷非织造布的性能

Properties forecasting of melt-blown nonwoven fabric by PSVM

用临近支持向量机(PSVM)建立了熔喷非织造布热空气温度、接收矩离、挤出量等3个工艺参数与过滤效率和透气量两大性能之间的关系预测模型,并用每组条件下两大性能的平均值对模型进行了验证.结果表明:过滤效率和透气量预测的拟合误差分别为0.122和38.49,而预测值与实测平均值之间的相关系数R都接近于1.经对比认为,PSVM的拟合效果非常理想.在此基础上通过计算机模拟方法得到最优工艺参数为:过滤效率最优工艺条件为温度270℃,DCD 15～19 cm,定量100 g/m2;透气量最优工艺条件为温度270～300℃,DCD 20 cm,定量160 g/m2.

By using proximal support vector machines（PSVM）,a forecasting model is established between the three process parameters（temperature,DCD and quantum）and the two properties（filtration efficiency and air permeability）of nonwoven fabric,and the mean values of the properties under each conditions are used to test the model,which indicates that the imitative errors of filtration efficiency is 0.122 and that of the air permeability is 38.49,and the correlation coefficients are both close to 1 for the two properties between the predicted values and the values from the experiment.As a result,conclusions are drawn that PSVM is ideal for the imitative effect of the experiment.Based on this,the optimum technological parameters are around temperature 270 ℃,DCD 15-19 cm and quantum 100 g/m^2 for the filtration efficiency and temperature 270-300 ℃,DCD 20 cm and quantum 160 g/m^2 for the air permeability attained by computer simulation.