Performance comparison of ANFIS models with different input spatial partitioning methods for predicting temperature and flow fields in a solar chimney

This study aimed to compare prediction performance between ANFIS models which use different input partitioning methods in case of predict temperature and flow field in a solar chimney. In this study, a fixed geometry solar chimney is simulated by CFD software to create dataset for ANFIS models. Three different ANFIS models consist of ANFIS1, ANFIS2, ANFIS3 was trained to predict temperature and flow field in the solar chimney and their prediction performances are compared and analyzed to figure out which model is the best model to predict temperature and flow field in the solar chimney. The result shows that the ANFIS2 model is the best model with RMSE for T, V prediction equal to 0.94, 0.0202 for the training set and 0.97, 0.0218 for the testing set. Keywords: ANFIS, Input Partitioning method, Solar Chimney, CFD, Machine Learning.