基于AdaBoost_SVM的轧机的状态评估

为了解决大型轧机设备的早期状态评估难的问题,针对样本数量较少和质量不佳时ANN表现出的过学习和欠学习的现象,及传统的SVM多用于二分类的问题,提出了一种基于AdaBoost_SVM算法的轧机状态评估方法。通过AdaBoost算法连接多个SVM弱分类器,从而得到分类准确率更高的强分类器AdaBoost SVM模型。该算法在轧机数据集上进行了测试,并且与传统的ANN算法、SVM算法进行了比较,实验结果表明AdaBoost_SVM算法具有更好的分类精度。

State Evaluation of Mill Based on AdaBoost_SVM

In order to solve the problem of large mill equipment＇s early evaluation, according to the 1： of over-learning and owe-learning about ANN （artificial neural network） when the sample amount is less is not so good,the problem that traditional SVM （support vector machine） is usually used in binary-class, a state evaluation method of rolling mill based on AdaBoost_SVM algorithm. Through connecting multiple classifiers by AdaBoost algorithm,we can get a strong AdaBoost SVM classifier model with higher classific henomenon and quality put forward weak SVM ation aecu- racy. The algorithm were tested on the mill data sets, compared with the traditional ANN algorithm and the SVM al- gorithm, the experimental results show that AdaBoost_SVM algorithm has better classification accuracy.