Hi,
Is it possible to extrapolate/predict wind speed data using artificial intelligence technique such as ANFIS, ANN, SVR for a location where no wind speed data is available and given that I have long time series wind speed data of some neighboring locations?
I know that based on neighboring location's meteorological data wind speed of a target station/location can be predict if target locations has recorded wind speed data.Based on my knowledge, AI techniques need to train and test using input and output data set and output data should be target parameter such as wind speed in my case.
And honorable experts, do you think that AI techniques has extrapolation capabilities (As my case) ?
There ought to be a way to achieve your goal.
Maybe you can find another large area of a similar topology for which you have the data. You learn how the wind behaves in this area regarding its neighbouring areas (for which you also have the data). Then, you have a model of wind that you may reuse for the area you don't have any data. The validity of this depends on the level of accuracy you want, and the level of similarity between your target area and the one you learn from.
This is just the first idea that popped in my mind, it's probably flawed somewhere.
@ Mr. Jérémy Boes , Thank you for your suggestions. Unfortunately, I did not understand clearly your opinion. Would you mind to elaborate your explanation ?
@ Zeashan Khan, Thank you for your reply. I want to predict monthly average wind speed and i have 4 location's daily as well as monthly average data of 10 years (132 months). As I said earlier, I want to predict monthly wind speed of neighboring location that has no wind data recorded. And want to use ANFIS. I think your answers is very short to get a clear idea about it. would you mind to write elaborately please ? And would you mind to suggest some reference works/papers that solved such type of problem ? Thank you again.
First broad answer:
You'll get the first best solution IF you have historical data for wind for the location/area of your interest.
IF your first best solution is not possible, you can 'approximate' your first best solution by your second best solution as suggested by Mr. Boes, using the similar data from another location that has similar characteristics as the location that you have interest.
To further clarify his point, say the Location-A is the location that you have interest to do your AI method in predicting wind speed (hence wind output). But since there is NO historical wind speed data for Location-A. But, you have wind speed data for Location-B which has similar geographic patterns and characteristics as Location-A. Then, you can use the historical data from Location-B and assume that these data also apply to Location-A. That approach will give you second best solution.
Clear??
@ Jeremy Lin, Thanks. Your suggestions seems good. What you mean by zero wind speed? where is that?
- Badges
- Science topic