Design of solar air conditioning system integrated with photovoltaic panels and thermoelectric coolers: Experimental analysis and machine learning modeling by random vector functional link coupled with white whale optimization

This research introduces a microclimate solar cooling system to enhance human thermal comfort and reduce electrical grid energy-based consumption. A novel solar photovoltaic thermoelectric air conditioner (SPVTEAC) for local air conditioning of a 1.0 m³ compartment was experimentally examined under several interior cooling loads. In this system, PV modules generate electric power, which is directly utilized to power the SPVTEAC and lead acid batteries for the self-service night operation of the hybrid system. The cooling performance of the SPVTEAC is analyzed in terms of the input PV current, cooling capacity, coefficient of the performance (COP), and mean air-cooled room temperature. Moreover, an optimized artificial intelligent modeling is also developed to model and validate the performance of SPVTEAC. The optimized modeling consists of an advanced random vector functional link network (RVFLN) coupled with a white whale optimizer (WWO) to propose fine-tuned fast modeling and explore the optimal parameters of the RVFLN. The experimentations were carried out for the SPVTEAC at hot climatic conditions of Sohag, Egypt, and their results were compared with the predicted RVFLN-WWO and stand-alone RVFLN findings. Furthermore, the robustness of the two models is evaluated using six statistical measures. The experimental outcomes indicated that the SPVTEAC could maintain a daily average air compartment temperature of 27.50 °C with COP of 1.14 when the interior thermal load, average air intake temperature, and air volume flowrate were 195 W, 36.50 °C, and 14.40 m³/h, respectively. Also, the statistical transactions revealed a unique fit between the experimental results and the predicted findings obtained by the proposed artificial intelligent model. It is asserted that the determination coefficient (R²) of the predicted PV electric current for the RVFLN-WWO and stand-alone RVFLN was 1.00 and 0.9151, respectively.