Open AccessResearch on short-term load forecasting based on feature similarity using PSO algorithm
Author(s) -
Xinquan Wei,
Weiyan Zheng,
Xiangjun Duan
Publication year - 2020
Publication title -
iop conference series. earth and environmental science
Language(s) - English
Resource type - Journals
eISSN - 1755-1307
pISSN - 1755-1315
DOI - 10.1088/1755-1315/510/2/022054
Subject(s) - particle swarm optimization , feature (linguistics) , term (time) , cluster analysis , similarity (geometry) , data mining , dimension (graph theory) , algorithm , computer science , residual , relation (database) , mathematical optimization , pattern recognition (psychology) , artificial intelligence , mathematics , linguistics , philosophy , physics , quantum mechanics , pure mathematics , image (mathematics)
In this paper, a short-term power load forecasting model based on feature similarity was proposed. The model can comprehensively consider the short-term load forecasting under the condition of multiple factors, such as meteorological information, holiday/workday information and other factors in a unified framework. Different values of the same characteristic have great influence on load forecasting. Therefore, hierarchical clustering algorithm is used to analyse the value of each feature dimension. Feature distance is used as feature mapping value to establish feature mapping relation table. Because different factors have different weights for load forecasting, a weighted feature similarity measurement strategy is designed. Taking the minimum residual sum of squares as the optimization objective, particle swarm optimization algorithm is used to solve the optimal characteristic weight. The validity of the model and the accuracy of load forecasting are verified by comparing the numerical simulation with the existing models.