Prediction and Optimization of Hourly and Cumulative Yield for Solar Still Using ANN-GA and ANN-PSO

Inadequate levels of potable water have led to worldwide freshwater scarcity being a major concern. Conventional desalination methods that mostly use fossil fuel energy sources are energy-intensive and emit harmful greenhouse gases (GHG), resulting in an adverse impact on the environment. Solar desalination is a process that uses solar thermal energy to remove unwanted salt and pollutants from saline or brackish water. It has emerged as a promising alternative technology to produce fresh water by harnessing renewable solar energy. A comprehensive investigation into the numerical optimization of solar still yield is presented in this study by incorporating state-of-the-art numerical methods. The investigation focuses on utilizing an artificial neural network (ANN) as the primary element, which is then integrated with the Genetic Algorithm and particle swarm optimisation (PSO). In order to accurately predict and subsequently optimize the performance of the system under a range of relevant operating conditions, the ANN-GA and ANN-PSO models constitute an essential component. Various meteorological parameters, namely, solar radiation, ambient temperature, ambient relative humidity, and wind velocity, are recorded in situ on a diurnal, monthly, and yearly basis in Shillong, Meghalaya, India (25.57° N, 91.89° E). Solar radiation, ambient temperature, glass temperature, and water temperature are considered input parameters for numerical analysis. The output parameters, which are to be optimized, include hourly yield, cumulative yield, and efficiency of the solar still after optimization. MATLAB toolbox is used for ANN-GA. However, codes are used to execute the analysis using the ANN-PSO method. The results show that the ANN-PSO has a better performance as compared to ANN-GA. The value of R = 0.99676, MSE = 0.00001427 for ANN-PSO, and the value of R = 0.69554, MSE = 0.0015. Hence, we can say that the ANN-PSO and ANN-Ga have the potential to predict the performance of a solar still.