直推式支持向量机(support vector machine,SVM)是基于已知样本建立对特定的未知样本进行有效识别的理论框架,与归纳式支持向量机相比,前者更经济、分类效果更佳。然而,直推式支持向量机的致命缺点是需要占用大量的训练时间,为此,提出了基于增量学习的支推式支持向量机训练算法,即把当前迭代训练得到的支持向量样本与新赋予类别标签的部分测试样本作为训练样本集参与下一次的迭代训目的是通过减少训练样本的数量以节约训练时间。同时,为确保算法的收敛性及分类准确率,在训练过程中引入了成对标注及错误回溯处理。实际的状态判别结果证明了该方法的有效性。 The support vector machine (SVM) is a theoretical framework based on known samples to effectively identify certain unknown samples. Compared with the induction SVM, the former is more economical and the classification is more effective. However, the deadly disadvantage of the direct support vector machine is that it takes a lot of training time. Therefore, this paper proposes a support vector machine training algorithm based on incremental learning, that is, the support vector samples obtained by the current iterative training and the new Part of the test label assigned to the category label as a training sample set involved in the next iterative training program is to reduce the training sample by reducing the number of training time. In the meantime, in order to ensure the convergence of the algorithm and the accuracy of the classification, pairwise annotation and error backtracking are introduced in the training process. The actual state discrimination results prove the effectiveness of this method.