基于相似性样本的LSSVM短期风速和风功率预测研究

风速具有较大的随机波动性，影响电网的稳定性，良好的风速预测是解决风电并网问题的关键。为了提高风速预测的精确性，首先对风速数据进行相似性样本的提取，采用分段线性化的搜索方法，求出各小段风速的斜率与长度所占的比重，继而找出相似性距离最小的曲线簇。并以此作为训练样本，采用最小二乘支持向量机（LSSVM）模型对风速进行预测。预测结果表明，采用风速的相似曲线簇进行LSSVM模型训练所得的风速和风电功率预测结果更优。

Short-Term LSSVM Wind Speed and Wind Power Prediction Based on Similar Samples

Wind speed is characteristic of large stochastic volatility, which affects the stability of the grid connected with it. Good prediction of the wind speed is the key to solve the integration of wind power with grid. In order to improve the accuracy of the wind speed prediction, first of all, the similarity sample of wind speed data is extracted, then using pieeewise linear search method the slope of each subparagraph wind speed and the proportion of length are calculated. Furthermore, the minimum similarity distance curve cluster is obtained. Taking it as a training sample, and using the least squares support vector machine （LSSVM） model, we can predict the wind speed. The forecast results show that .it is better to use similar wind speed curve clusters for LSSVM model training.