Optimizing the Solar-Driven Photocatalytic Activity of Er-Doped BiFeO3 Using Support Vector Regression and Particle Swarm Optimization

The welfare and security of future generations are progressively jeopardised by the swiftly expanding global population and the ecological consequences of industrialization-primary catalysts of climate change and the persistent energy crises. Given that fossil fuels provide more than 80% of global energy consumption, an immediate transition to cleaner alternatives is imperative. Hydrogen (H<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf>), as a renewable energy carrier, is a viable answer, especially when generated through solar photocatalysis. In this context, BiFeO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</inf> (bismuth ferrite) serves as an effective and environmentally benign photocatalyst for solar-driven hydrogen production. This paper presents a thorough normalisation strategy that incorporates square root, Box-Cox, Z-score, and min-max methodologies to mitigate skewness in a dataset comprising 24 photospheric magnetic field variables, hence enhancing system optimisation. A new framework integrating a nonlinear SVM-PSO algorithm was created for business optimisation. This hybrid SVM-PSO model was utilised for production optimisation, revenue forecasting, and project scheduling. Of the four assessed models, the proposed SVM-PSO exhibited the highest accuracy at 97.42%. These findings underscore the significance of sophisticated machine learning and data processing methodologies in enhancing clean energy applications, namely in solar photocatalysis utilising BiFeO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">3</inf>.