Hallo. If you have measured data, using machine learning models can be highly effective. Numerous studies have shown that these models can predict various types of radiation more accurately than traditionally used models. In my research, I used Random Forest Regression (RFR), Support Vector Machines (SVM), and Multiple Linear Regression (MLR) models to model net radiation, achieving very good results.

All of them are accurate for usage in CSP. Personally, Comsol is highly useful for me but all of those should be fine.