数据-模型混合驱动的风电场聚合等值建模方法

Hybrid Data-model-driven Aggregation Equivalent Modeling Method for Wind Farm

目前风电场等值建模方法主要针对全信息的情况,未考虑实际风电运行信息缺失的现状,且无法计及同群机组动态特性的差异。针对这两个问题,提出基于神经网络匹配算法的机组级信息获取方法,以克服实际场站机组级实时出力信息缺失难题;根据风电机组全风速的响应特性,分析确定分群指标,提出了风电场最少等值机表征原理,发现了传统等值方法的误差来源于对同群机组动态特性差异的忽略,进而提出了基于等值机动态行为校正的两机聚合等值建模方法。结果表明,提出的方法在不同的风速场景、电压跌落与故障持续时间下,均能够很好地模拟故障穿越响应特性。

Current equivalent modeling methods for wind farms are mainly aimed at the situation of information holography,without considering the lack of actual wind power operation information,and cannot take into account the dynamic characteristics differences of wind turbines within the same cluster.To solve these two problems,a unit-level information acquisition method based on the neural network matching algorithm is proposed,which overcomes the problem of lack of real-time output information at the unit level in actual stations.According to the response characteristics of the full wind speed of the wind turbine,the clustering index is analyzed and determined.The characterization principle of the minimum equivalent machine of the wind farm is proposed,and it is found that the error of the traditional equivalent method results from ignoring the dynamic characteristic difference in the same cluster of units.Therefore,a two-machine aggregation equivalent modeling method based on the dynamic behavior correction of the equivalent units is proposed.The results show that the proposed method can well simulate the response characteristics during the fault ride-through processes in different wind speed scenarios,and with different voltage dip and fault-duration time.