Abstract: The grey system forecasting model，neural network forecasting model and SVM（Support Vector Machine） forecasting model are proposed in this paper.Taking library budge of a library from year of 1996 to 2003 as a study case，the forecasting results are gotten by three methods.From the forecasting results，it is concluded that the accuracy of the SVM forecasting method is higher.Analyzing the characteristic of combining forecasting method，based on grey system forecasting model，neural network forecasting model and SVM，the linear combining forecasting model and SVM combining forecasting model are set up.Compared with single prediction methods and linear combining forecasting method，the accuracy of the SVM combining forecasting method is higher.

摘要： 对灰色、神经网络和SVM（支持向量机）的3个预测模型进行了研究，以某图书馆1996年～2003年图书文献总经费为例，对图书文献总经费进行了预测，经过比较，SVM的预测方法精度较高。在分析组合预测特性的基础上，提出了对灰色系统、神经网络和SVM三种预测方法结果进行了线性组合预测方法和SVM的组合预测方法。与单一预测方法结果和线性组合预测进行对比，SVM组合预测方法比较精确。