This thesis investigates the optimisation of fixed-tilt bifacial agricultural photovoltaic (agri-PV) orchard systems using a multi-objective approach, focusing on maximising both crop Photosynthetically Active Radiation (PAR) and PV radiation. Due to the complexity of modelling agri-PV systems, advanced light simulation tools like Radiance, a backward ray-tracing software, were integrated into the study. Although effective, this software introduces significant computational intensity and stochastic behavior.

The primary aims of this thesis were to develop an optimisation algorithm for such irradiation models and to optimise the system by maintaining separate objectives for crops and PV modules. This separation eliminates inter-connectivity between the conflicting metrics during the optimisation process. Additionally, a single-objective optimisation was conducted by summing the two conflicting objectives to assess the impact of this separation.

The research concluded that Bayesian optimisation with Gaussian Processes is most suitable due to its efficiency and ability to handle noise. Applied to an apple orchard case study in northern Italy, the algorithm demonstrated significant advantages over classical Radiance-based methods, achieving high accuracy in a fraction of the time. Results indicated that multi-objective optimisation offers more robust and informative solutions compared to single-objective optimisation.
In conclusion, this thesis enhances the understanding and optimisation of agri-PV systems, enabling quicker, more accurate analyses and providing practical solutions for farmers and other users.