电力系统暂态稳定概率评估方法

提出了一种基于蒙特卡罗-支持向量机的电力系统暂态稳定概率评估方法。首先构建了一组包含电力系统稳定和故障信息的原始特征，经特征选择降维后作为支持向量机的输入，在训练集上进行10折交叉验证，研究了4种支持向量机，其中径向基核支持向量机具有优良的评估性能；然后采用非序贯蒙特卡罗模拟方法选择随机因素，径向基核支持向量机加速暂态稳定评估过程，利用累计分类结果计算电力系统暂态不稳定概率。新英格兰39节点测试系统算例表明，该方法能大幅减少模拟时间，满足暂态稳定概率评估的精度要求。

Approach to Assess Power System Transient Stability Probability

Based on Monte Carlo-support vector machine (SVM), an approach to assess power system transient stability probability is proposed. A set of high-dimension features containing original features of power system stability and fault information is constructed and after the feature selection and dimension reduction the set is taken as the input of SVM and 10-fold crossvalidation are conducted with training set; four kinds of SVMs are researched and research results show that the radial base kernel SVM possesses good assessment performance; by use on non-sequential Monte Carlo simulation, the stochastic factors are chosen and the radial base kernel SVM is adopted to accelerate the assessment of transient stability, meanwhile, the transient instability probability of power system is calculated by the accumulation of classification results. Calculation results of New England 39-bus test system show that using the proposed method the simulation time can be evidently saved while the requirement to the accuracy of transient stability probability assessment can be satisfied.