摘要
介绍了基于稀疏贝叶斯学习理论的模式识别技术相关向量机及其分类器,在此基础上构建了电力系统暂态稳定评估模型。以EPRI36电力系统暂态稳定仿真数据为例,在相同的数据输入和相同的仿真环境下同时构建相关向量机和支持向量机2种暂态稳定评估模型。仿真预测计算显示,作为一种全新的概率学习模型,相关向量机不仅得到了比支持向量机更高的预测精确度,而且还能得到支持向量机无法完成的概率性预测和更高的稀疏性计算。
The RVM(Relevance Vector Machine) based on the pattern recognition of sparse Bayesian learning and its classification are introduced ,based on which a power system TSA(Transient Stability Assessment) model is constructed. Taking EPRI36 system as an example for TSA simulation,the RVM and the SVM (Support Vector Machine) are compared under the same conditions with same input data. Results show that,as a model of probabilistic learning,RVM offers superior prediction accuracy,and results in the probabilistic prediction and higher sparsity.
出处
《电力自动化设备》
EI
CSCD
北大核心
2009年第9期36-40,共5页
Electric Power Automation Equipment
关键词
概率学习
贝叶斯理论
相关向量机
支持向量机
暂态稳定评估
probabilistic learning
Bayesian theorem
relevance vector machine
support vector machine
transient stability assessment