A hybrid adaptive neuro-fuzzy inference system integrated with equilibrium optimizer algorithm for predicting the energetic performance of solar dish collector

Solar energy exploitation has a vital role to fulfill sustainability and decrease the usage of non-renewable energy resources. Solar parabolic dish collector (SPDC) is an effective alternative to fossil fuels due to its high efficiency. Nevertheless, performance prediction, optimization, and working fluid selection of SPDCs are highly complex problems and need complicated calculations and/or costly time-consuming experiments. Artificial intelligence-based algorithms have been proven to be beneficial in modeling different solar systems. Therefore, this study proposes an improved method to predict the thermal performance parameters of SPDC with a cylindrical cavity receiver using a modified algorithm of the adaptive neurofuzzy inference system (ANFIS) integrated with equilibrium optimizer (EO). In the developed algorithm, EO is employed as a new metaheuristic approach to enhance the prediction accuracy of ANFIS via determining the optimal values of ANFIS parameters. To evaluate the performance of the developed method, ANFIS-EO is compared with ANFIS and the conventional artificial neural network. The three models were applied to compare and predict the temperature difference of working fluid, heat gain, and energy efficiency of cylindrical receiver for SPDC operating with two different solar working fluids, namely, multi-wall carbon nanotubes/thermal oil nanofluid and pure thermal oil. Moreover, five statistical criteria are utilized to evaluate the performance of the investigated algorithms. The statistical performance results showed that the ANFIS-EO technique had the best prediction accuracy among the three models, and can be regarded as a powerful optimization tool for predicting the energetic performance of SPDC. The predicted results obtained by the ANFIS-EO have an excellent determination coefficient of 0.99999 for all predicted responses. (c) 2021 Elsevier Ltd. All rights reserved.