Long-term Wind Speed Series Estimation Method for Islands Using Spatio-temporal Correlation

Adequate historical wind speed data is a prerequisite for island microgrid planning. Therefore, to address the problem of lack of historical wind speed data of the island to be planned, a method is proposed to estimate the long-term wind speed series of the target island using the spatio-temporal correlation of wind speeds of surrounding islands. Firstly, the time series intervals of the wind speed series of the surrounding islands are divided adaptively by utilizing the sliding window and the cloud model. Secondly, the similarity transfer relationships (STRs) between the wind speed series of each segment of the surrounding islands are matched according to the cosine similarity of the numerical features of the wind speed cloud model in each time series interval. Finally, considering the STR and the spatial location of the islands, the influence of each STR on estimating the wind speed series of the target island is expressed in terms of weights. The long-term wind speed series of the target island are estimated based on each STR and its weight subsequently. The results show that, compared with the method of using Pearson correlation coefficient (PCC) to calculate the correlation between the wind speed series of each day and estimate the long-term wind speed series of the island, the mean absolute error, the root mean squared error and the PCC between the estimated results obtained by the proposed method and the actual series are improved by about 7.31%, 17.98%, and 0.46%, respectively. The proposed method can achieve high accuracy in estimating the long-term wind speed series of islands. The paper can provide a reference for the wind speed prediction of islands in the absence of historical wind speed data. © 2023 Science Press. All rights reserved.