A New MCP Method of Predicting Long-term Wind Speed with Height Error Revision

In recent years, modern wind turbine generators have grown rapidly and wind power plants have been established, delivering clean and inexhaustible energy. Therefore, the need for effective methods to evaluate wind power. Based on the fact that wind field has some degree of spatial correlation, measure-correlate-predict (MCP) algorithms can use concurrent data from target sites and a nearby reference site to predict the wind resource at target sites for wind power development. During last 15 years, over a dozen of MCP methods have been established, which differ in terms of overall approach, model definition, use of direction sectors, length of data. There are linear regression model, composite of wind speeds at two-site model, vector regression method, composite of standard deviations of two datasets and so on.But MCP algorithms mentioned above can only predict wind speed of target site with the same altitude. If the target site is higher or lower than the reference site too much, the result will be unreliable. So a new MCP method with height error revision is proposed based on data of two wind measurements, including six-layer wind data in one year. The fitted equations of Weibull parameters k and c as the function of height have been derived. By means of fitted equations, the relationship between winds of high and low altitude can be formulated. So, a method for error reduction is presented.At last, a set of performance comparison are carried out. The coefficient of correlation, the mean speed, the wind distribution and the correct annual energy production are selected as metrics at the target site, and a sample wind turbine power curve is analyzed. The mean and standard deviation of those estimates are used to characterize results. Results indicate that the new MCP method with height error revision work much better than previous ones.