Augmentation and prediction of wick solar still productivity using artificial neural network integrated with tree–seed algorithm

This study introduces a modified artificial neural network (ANN) model based on the tree–seed algorithm (ANN-TSA) to predict the freshwater yield of conventional and developed wick solar stills. The proposed method depends on improving the performance of the ANN through finding the optimal weights of the neurons (elementary units in an ANN) using the TSA. The use of developed wick solar still (DWSS) with copper basin results in increasing the freshwater productivity by about 50% compared with that of conventional wick solar still (CWSS) with steel basin. Then, the proposed ANN-TSA method is utilized to predict the hourly productivity (HP) of CWSS with steel basin and DWSS with copper basin. The real recorded data of the system were used to train the developed models. The predicted HP results of the CWSS and DWSS using ANN-TSA as well as ANN were compared with the experimental results obtained. The present study proves that ANN-TSA can be used as an effective tool to predict the HP of the CWSS and DWSS better than the ANN based on different statistical criteria (R2, RMSE, MRE, and MAE).