Enhancing thermal performance and modeling prediction of developed pyramid solar still utilizing a modified random vector functional link

This study introduces a modified random vector functional link (RVFL) as an alternative prediction method for the thermal performance of a developed pyramid solar still (DPSS), which consists of a copper basin integrated with graphite nanofluids. Experimental work is performed to investigate the performance enhancement of the DPSS. The real experimental data are recorded and utilized to build the proposed prediction models based on artificial neural network (ANN). Four ANN prediction models are presented, evaluated and compared to strength the prediction of the thermal performance of the DPSS. Different six statistical criteria are used to determine the optimal prediction model to be implemented in the prediction of the hourly freshwater (HF) and instantaneous energy efficiency (IEE) of the DPSS. From the experimental investigation, the use of proposed DPSS has total freshwater productivity of 5.26 L/m². The performed comparative study shows that the proposed RVFL approach tunned by firefly algorithm (FA), called FA-RVFL, is of optimal performance to be the best model among the investigated prediction models. The proposed FA-RVFL model is characterized by a determination coefficient of 0.981 and 0.999 and regression values of 0.996 and 0.999, respectively for the total data sets of HF and IEE. The present study shows the efficiency of proposed DPSS to enhance the freshwater capacity. Besides, it proves that FA-RVFL can be used as an effective tool to predict the thermal performance of the solar stills compared to the other models with no need for further experiments, thus saving financial expenses, effort, and time.