Properties forecasting of melt-blown nonwoven fabric by PSVM

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/m2 for the filtration efficiency and temperature 270-300 ℃,DCD 20 cm and quantum 160 g/m2 for the air permeability attained by computer simulation.