Reliability Prediction of Power Grid Systems Based on Multi-kernel Support Vector Machine
To improve the reliability prediction performance of power grid system,multiple objective functions were con-structed and multi-kernel support vector machine algorithm was used to predict reliability.The reliability evaluation objec-tive functions were established by screening three key indicators of power supply availability,average outage time and average outage times of from the grid sample features,and the reliability index features are trained by multi-kernel support vector machine.The multi-kernel combination types of Gaussian kernel function,polynomial kernel function and Sigmoid kernel function were used to obtain the reliability prediction results of power grid system through multi-kernel support vec-tor machine solving of different objective functions,and then the better kernel function combination type for reliability prediction was determined.The results show that with reasonable selection of kernel function combination and grid relia-bility index,multi-kernel support vector machine has high prediction accuracy and good stability in power supply availa-bility,average outage time and average outage times,and the reliability prediction accuracy of the combination type of Gaussian kernel function and Sigmoid kernel function is the best,the combination type of Gaussian kernel function,poly-nomial kernel function and Sigmoid kernel function has the best predictive stability.
reliability of power grid systemmulti-kernel functionsupport vector machineobjective function
NETL
NSTL
万方数据
