Application of RBF Network and Neurofuzzy System for Prediction of Transient Heat Transfer Between Curvilinear Contacts

ABSTRACT Contact heat transfer between curvilinear contacts has gained much attention due to its highly complex nature. There is still no thorough study for satisfactory calculation of curvilinear contact heat transfer using theoretical or physics‐based models. On the other hand, using the existing data, machine learning provides a powerful technique, capable of forecasting the thermal contact conductance (TCC) for different materials. In the present research, heat transfer between contacting curvilinear surfaces is calculated using the inverse solution, then the efficiency of adaptive neurofuzzy inference system (ANFIS) and radial basis function (RBF) neural network in estimating the rate of heat transfer through these surfaces based on the input–output data acquired by the inverse method and based on experimental study is investigated. The basis of this analysis is based on the temperatures measured from sensors located in specific areas of two surfaces in contact in a laboratory setup. Using the measured temperatures and the inverse method, the transient temperature function between curvilinear contacts is calculated. For the ANFIS, different training functions are used to train the network, and the best function is selected. The results show that the ANFIS using the dsigmf input function and linear output function can predict the TCC through curvilinear contacts with the root mean square error of 0.007. However, the error of RBF neural networks is 0.0059 less than the error of the ANFIS model for thermal contact prediction. Compared with other published works, the proposed networks perform better than group method of data handling–type neural networks. The ANFIS and RBF‐type networks perform well in estimating the rate of contact heat transfer between curvilinear contacts. However, the RBF‐type neural networks perform best. This investigation presents efficient and accurate strategies for estimating TCC applicable in aerospace, automotive, and manufacturing industries.