Abstract: In order to predict the pore size and its distribution of needle-punched nonwovens effectively,twenty-four kinds of polypropylene needle-punched nonwovens were produced by varying needle density and needle depth. The pore size of these samples was measured via the bubble point method. Taking needle density and needle depth as the inputs, a model based on support vector machine was established to predict the pore size and its variation coefficient of needle-punched nonwovens. And the cross-validation method was used to optimize the structural parameters of the model. Results indicated that the prediction precision for pore size and its variation coefficient were both higher than 98%, and their CV value were both lower than 2%. The result of the subsequent verification experiment further confirmed the high prediction accuracy of the support vector machine model. In addition, the prediction performance of the support vector machine model is better than that of BP neural network model.

Key words: support vector machine; needle-punched nonwovens; pore size and its distribution; forecasting model