Assessing Long-Term Wind Conditions by Combining Different Measure-Correlate-Predict Algorithms

This paper significantly advanced the hybrid measure-correlate-predict (MCP) methodology, enabling it to account for the variations of both wind speed and direction. The advanced hybrid MCP method used the recorded data of multiple reference stations to estimate the long-term wind condition at the target wind plant site with greater accuracy than possible with data from a single reference station. The wind data was divided into different sectors according to the wind direction, and the MCP strategy was implemented for each wind sector separately. The applicability of the proposed hybrid strategy was investigated using four different MCP methods: (i) linear regression; (ii) variance ratio; (iii) artificial neural networks; and (iv) support vector regression. To implement the advanced hybrid MCP methodology, we used the hourly averaged wind data recorded at six stations in North Dakota between the years 2008 and 2010. The station Pillsbury was selected as the target plant site. The recorded data at the other five stations (Dazey, Galesbury, Hillsboro, Mayville, and Prosper) was used as reference station data. The best hybrid MCP strategy from different MCP algorithms and reference stations was investigated and selected from the 1,024 combinations. The accuracy of the hybrid MCP method was found to be highly sensitive to the combination of individual MCP algorithms and reference stations used. It was also observed that the best combination of MCP algorithms was strongly influenced by the length of the correlation period.