Professional Documents
Culture Documents
Final PPT - Selection of Solar Irradiance Method For Solar Thermal
Final PPT - Selection of Solar Irradiance Method For Solar Thermal
Final PPT - Selection of Solar Irradiance Method For Solar Thermal
Due to strong increase of solar power generation, the predictions of incoming solar energy are ,
acquiring more importance.
•Solar radiation is an important parameter in solar energy application due to generation from
photovoltaic (PV) is directly related to this parameter.
The accurate solar radiation forecasting produced for a solar thermal power plant improve the
profitability and plant utilization
Solar radiation varies nonlinearly due to atmospheric events such as cloudy weather, rain,
humudity etc. Therefore estimation of solar radiation is an attractive issue in solar energy field.
Selection Of Solar Irradiation Method
Statistical Models
Persistence
• Future Solar Irradiance (Xt+1) Depends On The Latest Measurement (Xt)
Xt+1 = Xt
Auto Regressive
• Notation : AR (p) - Where p is Order Of Model
X = Observed Value
𝑥ҧ = Mean Of Observed Values
𝑥ො = Predicted Value
Daily Solar Irradiance Isn’t A Suitable Forecasting Time Step To Be Used In Step Because The
Temporal Capacity Of Thermal Energy Storage Is Between Six And Nine Hours.
Half Daily Values Consist Of Two Values Of Accumulated Global Solar Irradiance For Each Day
Which Gives A High Detail Of The Dynamic Characteristic Of Solar Irradiance Divided By Solar
Noon.
Experimental Setup
For Each Station They Have Tested 10 Different Autoregressive Models AR(p) With Order
P= 1… 10.
The Different Configurations Of Neural Network Models NN(p)have Been Tested With
Different Input Vector Size From P = 1 ….10 .
ANFIS (Adaptive Neuro Fuzzy Inference System) Model Has Been Tested With Different
Vector Size P = 1…6
Optimum Models for each station
Solar Irradiation Measurement
Equipment Used : Pyranometer (Thermopile type)
Clearness Index Time Series Obtains Better Results In Models Of Lower Order Compare To Lost
Component.
NN And ANFIS Models Obtain Better Results From Lc Time Series. Except In Lerida Station
Where Clearness Index Time Series Is Easier To Simulate By Models.
Models Results
Conclusion
Selection Of The Final Model, In Case There Are Various Models With Similar Errors, Is Based
On The Selection Of The Model With Less Number Of Parameters.
Clearness Index Time Obtains Better Results In Models Of Lower Order Compared To Lost
Component. This Is An Indication That Clearness Index Time Series Have Less Complexity Than
Lost Component.