基于CK-means与CatBoost算法的电力系统暂态稳定评估

Power System Transient Stability Assessment Based on CK-means and CatBoost Algorithm

考虑到电力系统中,失稳数据样本远少于稳定数据样本,针对数据不平衡问题,论文首先提出了基于Canopy和K-means算法的数据分解方法,对稳定和失稳样本进行分解,缓解数据不平问题对评估效果的影响;其次利用CatBoost算法对分解后的数据进行暂态稳定评估;通过新英格兰10机39节点系统仿真算例,在准确率和召回率方面与其他发现进行了比较,论文方法都具有一定的优势;同时,考虑到同步相量测量装置(Phasor Measurement Unit,PMU)配置的实际情况,检验了算法在不同PMU安装数量情况下的评估效果,论文方法较传统方法,受PMU数量的减少影响小。

Considering that unstable data samples are far less than stable data samples in the power system,this paper first proposes a data decomposition method based on Canopy and K-means algorithm to decompose stable and unstable samples to alleviate the impact of data imbalance on the evaluation effect.Secondly,CatBoost algorithm is used to evaluate the transient stability of the decomposed data.A New England 10-machine 39-node system is simulated and compared with other findings in terms of accuracy and recall rate.The proposed method has certain advantages.At the same time,considering the actual configuration of synchronous phasor measurement unit(PMU),the evaluation effect of the algorithm under different PMU installation number is tested.Compared with the traditional method,the proposed method is less affected by the reduction of PMU number.