
Particle swarm optimisation aided least‐square support vector machine for load forecast with spikes
Author(s) -
Lin WheiMin,
Tu ChiaSheng,
Yang RenFu,
Tsai MingTang
Publication year - 2016
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.2015.0702
Subject(s) - particle swarm optimization , support vector machine , artificial neural network , radial basis function , computer science , convergence (economics) , process (computing) , mean squared error , reliability (semiconductor) , electrical load , function (biology) , data mining , mathematical optimization , artificial intelligence , machine learning , engineering , statistics , mathematics , power (physics) , physics , quantum mechanics , voltage , evolutionary biology , electrical engineering , economics , biology , economic growth , operating system
This study developed a load forecasting system for electric market participants. Combining the least‐square support vector machine (LSSVM) and particle swarm optimisation (PSO), a LSSVM_PSO was proposed for the solving process. The loads, temperature, and relative humidity of the Taipower system were collected in the Excel Database. Data mining techniques is used to discover meaningful patterns, with the PSO applied to adjust learning rates. The forecasting error can be reduced during the training process to improve both the accuracy and reliability, where even the spikes were nicely followed. The support vector regression, LSSVM, radial basis function neural network and the proposed LSSVM_PSO were all developed and compared to check the convergence and performance. Simulation results demonstrated the effectiveness of the proposed method in a price volatile environment.