Modeling method of support vector regression using multirate sampling

To improve robustness properties of support vector regression (SVR) in control system modeling, after obtaining the mathematical derivative results that the much closer normal data to outliers in the input space are, the less effect of outliers to SVR is, a modeling method using multirate sampling is proposed, which is based on on-line SVR. In this method, the multirate sampling technique is used to increase training data density, and a local-data-intensive sliding time window is built to reduce the training data number and eliminate outliers. Furthermore, the method is employed in multichannel electrohydraulic force servo synchronous loading system to predict the load output. Compared with the traditional single rate sampling method, the results indicate that this method has better robustness and prediction accuracy, and the prediction mean absolute percentage error is 0.66%, in which only two training data are added.