Applying support vector machine to predict the critical heat flux in concentric-tube open thermosiphon

The critical heat flux (CHF) prediction model based on support vector machine (SVM) is established, and has been applied to the experimental cases in concentric-tube open thermosiphon under a variety of operating conditions, together with a traditional back-propagation (BP) neural network model. In this process, the performance of SVM with different parameters of C, ε and δ2 has been investigated and the optimal parameter setting is obtained. Based on the experimental results, through the comparison to the prediction results of the two models, the mean square root error and the mean relative error from SVM model are found to be about 50% of those from BP network model. The simulation results demonstrate that the SVM method can achieve better accuracy and generalization than the traditional BP network model, and it is effective for the CHF prediction in concentric-tube open thermosiphon.

► Support vector machine (SVM) technology is used to predict critical heat flux (CHF).
► The comparison between SVM and BP methods has been made.
► The application cases demonstrate the effectiveness of the SVM method.
► The method based on SVM may be a new alternative predicting method of the CHF.