Open AccessVariational Bayesian inference for the probabilistic model of power load
Author(s) -
Dong Zijian,
Wang Yunpeng,
Zhao Jing
Publication year - 2014
Publication title -
iet generation, transmission and distribution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.92
H-Index - 110
eISSN - 1751-8695
pISSN - 1751-8687
DOI - 10.1049/iet-gtd.2013.0237
Subject(s) - probabilistic logic , inference , bayesian inference , computer science , bayesian probability , artificial intelligence , machine learning , mathematical optimization , mathematics , econometrics
It is an essential element to estimate the state of system loads in the emerging smart grid system. This study is focused on finding a distribution of power load profile through a Gaussian mixed model. The authors propose a modified variational Bayesian inference (VBI) method to set parameters of the model. Compared with other approximation techniques, such as maximum likelihood, expectation maximum (EM) algorithm, they show that the author's Bayesian approaches can solve the problem of data singularity and the problem of over‐fitting. Utilising the weights of components, their modified algorithm can dynamically remove some redundant components during iterations. The presented method in this study also reduces the computational complexity and thus improves the speed of iteration. They use practical data to examine EM algorithm, VBI algorithm and their improved VBI algorithm. The experiments verify the advantages of Bayesian inference approaches and the validity/reliability of the proposed algorithm. The performance of the algorithms is also demonstrated by the data from PJM Company.