Enhancing Predictive Accuracy of Renewable Energy Systems and Sustainable Architectural Design Using PSO Algorithm

Abstract

Listen

Translate article

This paper formulates and examines the approach of integrating PSO into the tune of DNNs for boosting the predictive capability in renewable energy systems and green building designs. The PSO method was then employed to select Key features such as; Solar Irradiance, Ambient Temperature, Panel Efficiency and Energy Output. The PSO-based feature selection resulted in significant enhancements across a set of four metrics, there was an improvement in accuracy from a previous 0.82 to 0.87, precision from the previous 0.78 to 0.83, as well as recall from the previous 0.76 to 0.81, and the F1-Score from a previous 0.77 to the current score of 0.82. Moreover, the RMSE values reduced from 0.27 to 0.23, and the AUC values enriched from 0.74 to 0.85. Thus, the results of the current study support PSO’s role in improving feature selection, which, in return, improves the predictive models of energy management. The paper presented emphasizes the possibility of the use of enhanced optimization algorithms in enhancing the best performing, less resource-intensive, and environmentally friendly energy solutions in architecture.