Purposes With the large-scale grid-connection of new energy and power electronic devices, the security and stability characteristics of new power systems are facing serious challenges, and data-driven methods provide feasible ideas for the establishment of transient stability assessment models. However, the black-box nature of the models themselves determines their decision-making basis being not known, which has become a key factor limiting their online application. Methods To address this issue, a power system transient stability assessment model based on improved XGBoost and its interpretability method were proposed. Results On one hand, by improving the learning ability of the machine learning model on the decision boundary samples, the transient stability assessment accuracy of the model is significantly improved under the premise of meeting the assessment speed requirement. On the other hand, in order to improve the interpretability of the assessment results, attribution analyses of the model’s assessment results are carried out based on the Shapley additivity principle from the perspectives of global features and local samples, separately. Conclusions The simulation results at IEEE 39 nodes and a provincial power grid show that the proposed method has higher assessment accuracy than that of traditional models, and also has good interpretability.
为深入分析关键特征对预测目标的影响,采用累积局部效应图(accumulated local effects plot, ALE)。该图通过局部效应分析,排除了特征间相互关联的干扰,可以更清晰地看到每个特征对目标变量的独立影响。此外,ALE图还能评估具有强相关性的变量对预测目标的联合影响,相较于部分依赖图,它提供了一种更为有效且无偏的分析方法。ALE图的构建过程涉及对预测变化的累积平均的计算,并将这些变化累加到特定的网格上,如公式(10)所示:
WUY X, HANX Q, NIUZ W,et al.Interpretable photovoltaic power interval prediction using multi-attention deep neural networks[J].Power System Technology,2024,48(7):2928-2939.
[3]
ZHANGZ G, KANGC Q.Challenges and prospects of building a new power system under the goal of carbon neutrality [J].Journal of CSEE,2022,42(8):2806-2819.
[4]
SUNH D, XUT, GUOQ,et al.Analysis of the “8.9” blackout in the UK and its enlightenment to China’s power grid [J].Journal of CSEE,2019,39(21):6183-6192.
[5]
HUANGD, CHENS Y, GRACEW W.Research on power system transient stability control technology based on wide-area measurement information [J].Power Generation Technology,2018,39(5):389-396.
[6]
ZHUL P, HILLD J.Data/Model jointly driven high-quality case generation for power system dynamic stability assessment[J].IEEE Transactions on Industrial Informatics,2022,18(8):5055-5066.
ZHANGL, ANJ, ZHOUY B.Transient stability assessment of power system based on temporal convolution and graph attention network[J].Automation of Electric Power Systems,2023,47(7):114-122.
[9]
ZHUL P, WENW J, LIJ Y,et al.Integrated data-driven power system transient stability monitoring and enhancement[J].IEEE Transactions on Power Systems,2024,39(1):1797-1809.
[10]
TANB, YANGJ, TANGY,et al .A deep imbalanced learning framework for transient stability assessment of power system[J].IEEE Access,2019,99(7):81759-81769.
LIUS K, DANGX, CUIZ Q,et al.Transient stability evaluation method for power systems with deep residual network considering class imbalance of samples[J].Smart Power,2024,52(1):116-123.
HUL T, WANGH Y, DANGR,et al.Imbalance correction method of transient stability assessment model based on gradient norm[J].Electric Power Automation Equipment,2024,44(4):158-163,177.
LIJ M, YANGH Y, YANL P,et al.Integrated assessment method for transient stability of power system under sample imbalance[J].Automation of Electric Power Systems,2021,45(10):34-41.
WUY X, HANX Q, NIUZ W,et al.Transient stability assessment method based on variable weight random forest and its interpretability analysis[J].Automation of Electric Power Systems,2023,47(14):93-104.
JIS L, LIJ F, DUT Y,et al.Survey on techniques,applications and security of machine learning interpretability[J].Journal of Computer Research and Development,2019,56(10):2071-2096.
[23]
DENGH T.Interpreting tree ensembles with intrees[J].International Journal of Data Science and Analytics,2019,7(4):277-28.
HANT S, CHENJ F, LIY H,et al.Study on interpretable surrogate model for power system stability evaluation machine learning[J].Proceedings of the CSEE,2020,40(13):4122-4130.
[26]
RENC, XUY, ZHANGR.An interpretable deep learning method for power system transient stability assessment via Tree Regularization[J].IEEE Transactions on Power Systems,2022,37(5):3359-3369.
[27]
CREMERJ L, KONSTANTELOSI, STRBACG.From optimization-based machine learning to interpretable security rules for operation [J].IEEE Transactions on Power Systems,2019,34(5):3826-3836.
[28]
ZHANGS, ZHANGD, QIAOJ,et al.Preventive control for power system transient security based on XGBoost and DCOPF with consideration of model interpretability[J].CSEE Journal of Power and Energy Systems,2021,7(2):279-294.
[29]
LIUC.A systematic approach for dynamic security assessment and the corresponding preventive control scheme based on decision trees[J].IEEE Transactions on Power Systems,2014,29(2):717-730.
YUEY J, WUM Y, WANGH J,et al.Short term photovoltaic power prediction based on CNN-GRU-ISSA-XGBoost[J].Journal of Nanjing University of Information Science & Technology,2024,16(2):231-238.
SUNH, SUNB S, GAOZ N,et al.Online identification for power system transient angle stability based on time-series spectral distribution characteristics of measured data[J].Power System Technology,2023,47(3):1107-1116.
SHAOM Y, WUJ Y, LIB Q,et al .Transient stability assessment of power system based on two-stage ensemble deep belief network[J].Power System Technology,2020,44(5):1776-1787.
ZHOUT, YANGJ, ZHANX P,et al.Data-driven method and interpretability analysis for transient voltage stability assessment[J].Power System Technology,2021,45(11):4416-4425.
ZHENY Z, RUANC.Reinforcement learning-based hybrid element heuristic transient voltage stability feature selection and Its interpretability[J].Power System Technology,2024,48(4):1519-1531.
LIP, DONGX J, MENGQ W.Transient stability assessment method for power system based on Fisher Score feature selection[J].Electric Power Automation Equipment,2023,43(7):117-123.
JIANGX X, XUJ, LIAOS Y,et al.An intelligent selection method for power system transient security key features based on enhanced evolutionary computation[J].Proceedings of the CSEE,2023,43(14):5358-5371.