And how to know which method is actually more reliable than other? We virtually cannot tell unless we see the measured data
A workmate of me have been working with GRP (Generalized Renewable Process) P. Jimenez. Please Check this link.
https://www.researchgate.net/publication/224056830_Generalized_Renewal_Process_as_an_Adaptive_Probabilistic_Model
Conference Paper Generalized Renewal Process as an Adaptive Probabilistic Model
Perhaps you would like to read the paper entitled A Review of Wind Power Forecasting Models and the paper entitled Wind Power Forecasting Error Distributions: An International Comparison
The answer probably depends on what you are trying to do and also what data you are using, and what forecasting method is being used. Probably the periodicity of the data (5imin, 15min, etc) matters too. I don't think it can be said that one model is right for all situations. For example, if you are forecasting wind power with a time-series based econometric model, then of course you make an assumption about the error structure and process of the model, but then also you can (hopefully) test these assumptions using within sample forecasts. You should also be clear about the criteria selection function (e.g. MSE forecast, etc). Once you get to this point you can start testing models of the error, and there are many possibilities within the general time-series literature (ARIMA, GARCH), etc. Hope that is helpful and let us know how you get on.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Energy
- Energy-Generating Resources
- Renewable Energy