基于相空间重构的支持向量回归机在振动数据预测中的应用

支持向量回归机是近年来发展起来的一种通用的机器学习方法,后来被广泛应用于预测领域。在对模型进行训练时,输入特征在很大程度上影响了预测的精度。所以对于特征的选择一直是人们所关注的问题。提出了一种基于相空间重构的支持向量回归机方法。该方法首先对时间序列进行相空间重构,然后利用重构的相空间中的相点作为特征输入,对模型进行训练。经实验验证,该方法能够根据时间序列内在规律,自适应的构造输入特征,提高预测结果的精度。

The support vector regression for vibration date forecast based on phase-space reconstruction

Support vector regression is a popular machine learning method that develops these years and has been widely used in the prediction field.The input characters greatly affect the prediction precision during training the models.Therefore,how to choose the character is always a problem that people have paid attention to.This paper puts forward a SVR method based on the phase space reconstruction.This method firstly reconstructs the phase space of the time series.Then the points of the phase space are used as the input character to train the models.The experiment shows that this method can self-adaptively form the input characters according to the inner principles of the time series,and improved the precision of prediction result.