基于风电机组聚类的风电场有功分层分配策略

HIERARCHICAL ACTIVE POWER DISTRIBUTION STRATEGY OF WIND FARMS BASED ON WIND TURBINES CLUSTERING

针对风电场传统有功功率平均分配策略造成风电机组运行损伤差异性大、机组调整和启停次数多等问题,提出基于风电机组聚类的降功率分层分配策略。考虑实际风电机组调控能力,设计基于鲸鱼优化算法(WOA)的改进极限学习机(ELM)风电机组功率预测模型。构建实际风速、实际发电功率和预测功率的三维特征矩阵,提出基于模拟退火遗传算法(SAGA)的改进模糊C均值(FCM)聚类模型。以预测功率平均值和标准差、最大降功率值确定机群和机组调度优先级序列,构造“机群-机组”的两层分配策略,减少调整机组和停机机组的数量、提升风电机组损伤均衡性。结合某风电场实际工况,设计限功率模式不停机与停机的分配策略验证方案,对比分析3种不同策略的分配效果,算例结果验证了提出的分配策略的有效性。

In view of the problems caused by the traditional average active power distribution strategy of wind farms,such as large differences in wind turbine operation damage,multiple adjustment and start-stop times,etc.,a hierarchical active power reduction distribution strategy based on wind turbine clustering is proposed in this paper.Considering the control ability of the actual wind turbines,a power prediction model of improved extreme learning machine(ELM)based on whale optimization algorithm(WOA)is designed.A three-dimensional characteristic matrix of the actual wind speed,actual generated power and predicted power of wind turbines is constructed,and an improved fuzzy C-means(FCM)clustering model based on simulated annealing genetic algorithm(SAGA)is proposed.The scheduling priority sequences of clusters and turbines are determined by the predicted power average value,standard deviation and maximum power reduction value,and a two-layer distribution strategy of“cluster-turbine”is constructed.The adjustment number of operating and shutdown turbines is reduced,and the degeneration balance among wind turbines is improved.Combined with the actual operation conditions of a wind farm,the verification scheme of the distribution strategy of non-stop and shutdown in the limited power mode is designed.The distribution effects of three different strategies are compared and analyzed.The results of the calculation example verify the effectiveness of the proposed distribution strategy.