Adaptive particle swarm optimization-based deep neural network for productivity enhancement of solar still

Abstract

Water is considered one of the most superabundant resources on the earth that covers 75% of the entire earth's surface, yet numerous countries face problem due to water shortage. Desalination is considered the most efficient process to overcome this rising clean water demand. Solar energy is considered one of the efficient and finest resources to refine brackish water. Therefore, this paper proposes a novel black widow particle swarm optimization-based deep neural network approach to enhance the water productivity from a solar still. The main intension of the proposed BWPSO-based DNN approach is to enhance the performances of DNN by employing BWPSO for optimal water production. Here, the optimal weight of the DNN is determined by utilizing the BWPSO algorithm. The solar still is incorporated with a straight tube and spiral tube solar water collector. In addition to this, the study based on solar still and their experimental analysis is carried out in Coimbatore city located in Tamil Nadu. The evaluation is conducted for various parameters, namely glass temperature, average evaporation temperature, inlet and outlet temperature, water temperature, air temperature, yield, solar intensity, wind velocity, RMSE, MAE, MRE, and ECR, to determine the effectiveness of the system. Also, comparative analysis is made and the evaluation results reveal that the proposed approach outperforms various other approaches.

Keywords: BWPSO; DNN; Galvanized sheet; Solar intensity; Solar still; Temperature; Water productivity.