Created a simulation of future wind speeds in Northeastern U.S And Southeastern Canada in R
Optimize the placement of renewable power plants to create a reliable green grid
Collected data was complied, cleaned, and pre-analyzed for use in the model
A gerneric sinsousldal model was created and fitted with prepocessed data
Residuals from the model are processed and used to create a simulation
Simulated results for all locations were analyzes to identify trends
Combined data across several large files From the NSRDB (National Solar Radiation Database)
Using a custom-made R script
Metadata
31 locations over 24 years
Predictors – Wind Speed
13,000,000 records
Preprocessing
Calculated daily average speed for each location
Captures the wind speed for each day
Differences are more meaningful
Randomly selected 15% of the data per location
Minimizing error isn’t the main objective
Scaled and Centered the data
Better Performance and normal error distribution
Visualized Data
Purpose
Model behavior across complex systems
Determine the locations that are most influential
Comparison of different scenarios to Determine the ideal locations
Need to capture randomness to find general trends
Not Anomalies
General Sinusoidal Model
Model Objective
Objective to select a combination of coeffects to minimize error
Simulation Objective
Appropriately model error to apply randomness to the simulation
f ̂(x)=f(x)+ϵ
The next 1000 days of wind speeds
Error distribution add the randomness
Random Error
Still simulated lower than expected wind speed
Predict peak wind days
Low Correlation Benefit
Locations with little correction could power each other
i.e. with less correlation, if one location has little wind, then another location is more likely to have stronger winds
Several locations are poorly correlated with most other locations
Can supply more locations power on still days
Southern Canada
Large body of water
Canadian Lowlands
North Coast of the Delaware Bay
Large body of water
Flat costal planes across the bay
Montréal
Small body of water
Middle of the large St. Lawrence River Valley
Combination of proximity to a body of water and geographical features
Locations with the least correlations also have the highest average and standard deviation of wind speed